Student Performance Analysis and Prediction in Classroom Learning: A Review of Educational Data Mining Studies

Educational and Information Technologies (2021) 26:205–240

https://doi.org/10.1007/s10639-020-10230-3

Student performance analysis and prediction

in classroom learning: A review of educational data
mining studies
Anupam Khan1 · Soumya K. Ghosh1
Received: 4 February 2020 / Accepted: 17 May 2020 / Published online: 1 July 2020
© Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Student performance modelling is one of the challenging and popular research top-
ics in educational data mining (EDM). Multiple factors influence the performance
in non-linear ways; thus making this field more attractive to the researchers. The
widespread availability of e ducational datasets further catalyse this interestingness,
especially in online learning. Although several EDM surveys are available in the lit-
erature, we could find only a few specific surveys on student performance analysis
and prediction. These specific surveys are limited in nature and primarily focus on
studies that try to identify possible predictor or model student performance. How-
ever, the previous works do not address the temporal aspect of prediction. Moreover,
we could not find any such specific survey which focuses only on classroom-based
education. In this paper, we present a systematic review of EDM studies on student
performance in classroom learning. It focuses on identifying the predictors, methods
used for such identification, time and aim of prediction. It is significantly the first
systematic survey of EDM studies that consider only classroom learning and focuses
on the temporal aspect as well. This paper presents a review of 140 studies in this
area. The meta-analysis indicates that the researchers achieve significant prediction
efficiency during the tenure of the course. However, performance prediction before
course commencement needs special attention.
Keywords Student performance · Classroom learning · Performance prediction ·

Literature review · Educational data mining
Anupam Khan
[email protected]
Soumya K. Ghosh
[email protected]
1 Department of Computer Science and Engineering, Indian Institute of Technology Kharagpur,

Kharagpur, West Bengal 721302, India
206 Educational and Information Technologies (2021) 26:205–240
1 Introduction
Student performance analysis and prediction are the two widely explored research
topics in education system literature. Although their objectives are different, the out-
comes of performance analysis significantly influence the prediction studies. The
statistical methods may not always be sufficient for establishing the association of
various factors with performance (Natek and Zwilling 2014). The use of sophisticated
algorithms may yield impressive knowledge that could help the educators as well as
students (Baker and Yacef 2009; Kumar et al. 2018; Liu et al. 2018b; Romero and
Ventura 2007). The advancement of data mining technologies has influenced many
researchers to investigate more profound insights into the knowledge dissemination
process. Some of them have applied data mining approaches for this purpose long ago
(Fausett and Elwasif 1994; Gedeon and Turner 1993). However, the number of such
studies are very less at that time. The growing availability of digital data captured
by several academic information management systems and educational software in
recent years catalyses this process to improve the quality of education (Aghabozorgi
et al. 2014; Asif et al. 2017; Baker 2014; Khanna et al. 2016; Livieris et al. 2018;
Loh and Sheng 2015; Ogor 2007; Wook et al. 2016). Interestingly, Baker and Inven-
tado (2014) estimated that all educational research would involve analytics and data
mining by 2022.
The researchers have earlier conducted multiple surveys on EDM studies. Some of
them consider a broader scope to outline multiple aspects of educational processes.
A few of them specifically focuses on student performance analysis and prediction
as well. Let us briefly introduce these surveys here. The first one was published more
than a decade ago in which Romero and Ventura (2007) consolidate the studies pub-
lished between 1995 and 2005 and outline the probable objectives of EDM. In 2010,
they published another exhaustive review (Romero and Ventura 2010) which con-
siders nearly 300 papers published between 1993 and 2009. It divides the research
area of EDM into eleven broad sub-areas, and student performance prediction is one
of them. The area of learning analytics has also gained momentum by that time.
It is essential to mention that two research communities, the international educa-
tional data mining society and the society for learning analytics research, initially
drive the analytical studies in education. Siemens and Baker (2012) have mentioned
the overlap of research interest between these two communities. In the same year,
Ferguson (2012) outlines the drivers and challenges behind learning analytics studies.
In 2013, Chrysafiadi and Virvou (2013) carried out a specialised review on student
modelling approaches. It may be the first of this kind that focuses on a specific area
of EDM. Pen̄a-Ayala (2014) has published another exhaustive survey in 2014, which
considers 240 EDM studies between 2010 and the first quarter of 2013. According
to this study, the number of works on student performance modelling is increasing
significantly since 2010. The researchers have additionally carried out specific EDM
survey on the psychology of learning and application of clustering approaches as well
(Koedinger et al. 2015; Dutt et al. 2017). Bakhshinategh et al. (2017) have probably
conducted the most recent review of EDM studies in which they mention that student
performance prediction can help the educators and administrators significantly.
Educational and Information Technologies (2021) 26:205–240 207
In addition to the broad survey on EDM studies, a few specific surveys on stu-
dent performance analysis and prediction are also available in the literature. Shahiri
et al. (2015) have published first such survey in 2015 to the best of our knowledge.
It primarily focuses on the predictor attributes and the prediction methods used in
39 studies published between 2002 and 2015. Later on, Khanna et al. (2016) out-
line the parameters used for predicting students performance in 25 research papers.
Furthermore, another review of 40 papers in 2017 provides an overview of predic-
tion accuracy in addition to the factors and methods used (Hu et al. 2017). Kumar
et al. (2017) conducted a similar survey with 16 papers in the same year with the pri-
mary focus on used predictors, methods and accuracy. In another two short reviews
(Shingari and Kumar 2018; Kumar et al. 2018), the authors outline the data mining
techniques used for predicting performance in 17 and 36 papers, respectively.
Although the researchers have applied EDM techniques in both traditional and
computer-based online education, the application in conventional education is com-
paratively less than the other alternatives available. The authors of two separate
EDM review papers (Pen̄a-Ayala 2014; Romero and Ventura 2010) therefore found
only a few studies on the traditional education system. Moreover, all specific EDM
surveys on student performance consider the studies on both online and traditional
classroom-based learning. Importantly, we could not find any EDM survey paper
which focuses explicitly on classroom-based education. The previous surveys on
student performance primarily focus on the predictors, methods and prediction effi-
ciency. However, none of them considers the temporal aspects. The prediction before
and after course commencement may serve a completely different purpose. There is
a pressing demand for next-term performance prediction nowadays to provide timely
and effective support (Backenköhler and Wolf 2017; Polyzou and Karypis 2019;
Shingari and Kumar 2018; Sweeney et al. 2016). Prior knowledge of performance
influencing factors can help in implementing such advanced prediction (Helal et al.
2018). Therefore, this survey primarily focuses on the following research questions:
– What are the factors influencing student performance in classroom learning?
– Which methods are used for finding these factors?
– Is it possible to predict before course commencement?
– Can we predict the actual grade or score?
In this paper, we present a systematic survey of student performance related EDM
studies with a specific focus on classroom learning. Various technologies like online
discussion forum, assessment system, assignment and learning tools are used in
classroom learning nowadays. This survey does not restrict to the studies only on
those educational systems that follow the traditional instruction delivery mechanism.
Instead, it considers all works which are related to the students of the traditional
education system. In this paper, we present a review of 140 relevant studies pub-
lished between 2000 and 2018. In order to carry out this survey, we have created a
taxonomy of research directions and categorise the relevant works accordingly. The
meta-analysis finally helps in identifying future research directions.
The anatomy of the paper is as follows: Section 2 elaborates the methodology
adopted for conducting this survey. Sections 3 and 4 present the meta-analysis of rel-
evant works. In Section 3, we first outline the existing studies of student performance
analysis and Section 4 later presents the studies on performance prediction before
and after course commencement separately. Section 5 discusses the observations and
recommendations for future work. Finally, Section 6 concludes the survey.
2 Methodology
This study has followed the recommendation provided by Cooper (1988) for deter-
mining the focus, goal, perspective, coverage, organization and audience of this
survey (refer Table 1). The goal here is to identify the gaps or central issues with
a particular focus on student performance analysis and prediction related literature.
This neutral survey is not an exhaustive one; rather, it represents the relevant literature
pivotal to the research questions framed. In this paper, Fig. 1 presents the methodol-
ogy adopted for conducting this survey. The following parts of this section mention
the detailed steps.
In a literature survey, a well-planned and well-executed search strategy is essen-
tial for finding every relevant piece of work (Shahiri et al. 2015). Therefore, we
have first created a taxonomy of research directions as a part of the survey strategy.
Figure 2 presents the defined taxonomy here. The shaded part of Fig. 2 is the pri-
mary focus of this survey. We have organised the paper in these research directions.
Once the taxonomy is defined, we have adopted a hybrid approach for searching
the relevant literature. The Google Scholar helps us in configuring an alert with
(“educational data mining” and “student performance”) string. Due to this config-
uration, Google Scholar regularly sends a list of recently published relevant papers
to our email since August 2015. We have additionally adopted the conventional
approach for searching relevant literature as well. In this approach, we search the
Google Scholar with the following search strings: (i) “educational data mining” and
“student performance”, (ii) “learning analytics” and “student performance”, (iii) “stu-
dent performance” and “teaching quality”, (iv) “student performance” and “domain
knowledge”. This approach facilitates us to get the studies published before August
2015 as well. Importantly, Google Scholar is selected here for better coverage of the
search space (Walters 2007).
The above hybrid search approach retrieves multiple articles which are filtered
systematically to obtain the list of selected articles. The recommendation of the
Table 1 Scope of the survey

Characteristics Scope
Focus Research Outcome

Goal Identification of central issues
Perspective Neutral representation
Coverage Representative, pivotal
Organisation Conceptual
Audience Specialised scholars
Fig. 1 Methodology adapted for this survey
Prisma Statement (Moher et al. 2009) helps in the selection process here. As an ini-
tial step, we have checked the title to identify studies related to student performance
analysis or prediction. This process helps us in identifying a total of 362 distinct
studies as a potential candidate for this survey. We read the abstract and keyword
Fig. 2 Taxonomy of research directions

mentioned in each one of them for further screening, and it filters out 65 irrelevant
studies. In the next step, we have read the full-text and tried to identify whether
the contribution of the study is pivotal to the research questions of this survey. This
process helps us in excluding the studies which are not related to the students of
traditional classroom-based education as well. We also discard those studies which
are not well-presented, contains unclear methodology, dataset or contribution. Due
to these criteria, 157 studies do not pass the eligibility test, and 140 relevant articles
finally get selected for this review. In the next step of the qualitative synthesis, we
add the shortlisted articles in Mendeley (Zaugg et al. 2011), a reference management
tool provided by Elsevier. During inclusion, we also classify each article with mul-
tiple tags. Each item in the defined taxonomy except those on the shaded region of
Fig. 2 are the tags used here. It helps us in grouping the articles further. Moreover,
we use the general note section of Mendeley to keep a track on the key findings, sig-
nificant contributions and limitations of each article which eventually facilitates us
to complete the meta-analysis.
3 Student performance analysis
The choice of predictors plays a vital role in any efficient prediction task. The fac-
tors affecting knowledge dissemination process, therefore, should be identified first
before predicting student performance (Helal et al. 2018). Furthermore, the decrease
in students’ success rate and frequent drop-outs are the primary concern of many
educational institutes nowadays. Prior analysis of student performance could help the
institutes in this regard (Mat et al. 2013). This section presents a meta-analysis of
existing studies which tries to identify and measure the influencing factors of student
performance. It first introduces the factors found in the literature and later on dis-
cusses the methods used for establishing the association of these factors with student
performance.
3.1 Influencing factor identification
Various factors impact the performance in non-linear ways; thus, their identification
for subsequent use as predictors of success is a complex problem (Zollanvari et al.
2017). The social and academic background of the student, their behaviour, previous
assessments, quality of teaching, previous knowledge on the course topics are a few
broad areas explored by the researchers. Here we present a meta-analysis of existing
literature. Besides this, Tables 2, 3, 4 and 5 summarise them in tabular format for
providing a comprehensive view to the readers.
3.1.1 Student background and behaviour
The exploration of student demographics, behaviour, academic and social back-

ground related factors are widely available in student performance analysis literature.
The influence of gender is one of the widely debated topics among them. Ahmed
et al. (2014) show that the student is more likely to be a male student in case of poor
Table 2 Student background and behaviour related studies
Study Explored predictor Method
Ahmed et al. (2014) Gender, residence type ARM

Al-Obeidat et al. (2017) Parent education and job CLA
Bayer et al. (2012) Communication skill, social behaviour SNA, CLA
Bendikson et al. (2011) Socio-economic status REG
Campagni et al. (2015) Order of examination taken CTR
Carter et al. (2017) Programming and social behaviours REG
Daud et al. (2017) Family expenditure and personal information CLA
Dvorak and Jia (2016) Study habits REG
Felisoni and Godoi (2018) Cell phone usage behaviour REG
Fernandes et al. (2018) Age, Social and academic background CLA
Garcı́a et al. (2007) Learning management system usage behaviour ARM
Gasevic et al. (2017) Learning strategies CTR
Gowda et al. (2013) Learning style REG
Gray et al. (2014) Psychometric factors CLA
Guruler et al. (2010) Registration category and family income CLA
Hart et al. (2017) Attitude, cognitive skills, and engagement REG
Hattie and Clinton (2012) Physical activity REG
Hassan and Rasiah (2011) Poverty level REG
Helal et al. (2018) Social background CTR
Ivančević et al. (2010) Spatial position in laboratory ARM, CTR
Mishra et al. (2014) Social, academic background, emotional skills CLA
Natek and Zwilling (2014) Past academic and social background CLA
Pal and Chaurasia (2017) Alcohol consumption CLA
Papamitsiou et al. (2014) Temporal behaviour of answering questions REG
Quadri and Kalyankar (2010) Scholarship status, first child CLA
Ramesh et al. (2013) Parent occupation REG
Romero et al. (2008) Activity in moodle CLA
Romero et al. (2013) Participation in online discussion forum SNA, CTR
Saarela and Kärkkäinen (2015) Learning capabilities REG, CTR
Saxena and Govil (2009) Parent occupation, student demographics CTR
Van Inwegen et al. (2015) Assignment tool usage pattern REG
Wang et al. (2016) Assignment tool usage pattern REG
Zacharis (2015) Activity in learning management system REG
Zimmermann et al. (2015) Performance in undergraduate level REG
ARM: Association rule mining, CLA: Classification, CTR: Clustering, REG: Regression, SNA: Social
network analysis
performance. In contrast, Quadri and Kalyankar (2010) did not find any influence of
gender on student performance. They instead find a significant impact of low family
income on performance and drop-out. A few other studies have also reported a similar
observation (Daud et al. 2017; Guruler et al. 2010; Hassan and Rasiah 2011). Inter-
Table 3 Assessment related studies
Ahmed et al. (2014) Internal and external assessment, attendance ARM

Al-Barrak and Al-Razgan (2016) Performance in other courses CLA
Beemer et al. (2018) Internal assessment REG
Bucos and Druagulescu (2018) Attendance, average score and credits, CLA
student activity
Buldu and Üçgün (2010) Previous failure ARM
Buniyamin et al. (2015) CGPA, past academic performance CLA
Chaturvedi and Ezeife (2013) Assignment marks ARM
Christian and Ayub (2014) CGPA CLA
Fernandes et al. (2017) Attendance CLA
Golding and Donaldson (2006) Assessment in first year courses REG
Guarin et al. (2015) Admission test score CLA
Helal et al. (2018) Internal assessment CTR
Huang and Fang (2013) CGPA, internal and external assessment REG
Jishan et al. (2015) CGPA, attendance and internal assessment CLA
Kamley et al. (2016) Performance in school ARM
Kaviyarasi and Balasubramanian (2018) Internal test scores CLA
Li et al. (2013) Score in first year mathematics REG
O’Connell et al. (2018) Performance in prior courses REG
Parack et al. (2012) Internal assessment ARM
Romero et al. (2008) Internal assessment data in moodle CLA
Superby et al. (2006) Attendance, external assessment CLA
Tair and El-Halees (2012) Performance in secondary education ARM, CLA
Xiong et al. (2014) Prerequisite skills REG
ARM: Association rule mining, CLA: Classification, CTR: Clustering, REG: Regression
estingly, the fact of whether the student is the first child of the parent also found to
be influential on performance (Quadri and Kalyankar 2010). Some studies have even
reported the possible impact of parent occupation (Al-Obeidat et al. 2017; Ramesh
et al. 2013), registration category (Guruler et al. 2010; Helal et al. 2018), past aca-
demic and socio-economic background of the student (Bendikson et al. 2011; Helal
et al. 2018; Natek and Zwilling 2014), neighbourhood, previous school and age (Fer-
nandes et al. 2017, 2018) as well. In contrast, Hattie and Clinton (2012) show that
physical activity does not influence student performance at all. However, Gray et al.
(2014) exhibit that psychometric factors like learner ability, personality, motivation,
and learning strategies are useful to identify the college student who is at risk of fail-
ure. In a similar study, Mishra et al. (2014) mention that the social and academic
integration and various emotional skills of the student are also capable of building a
performance prediction model.
Table 4 Teaching related studies
Abrami et al. (2007) SETE REG

Balam and Shannon (2010) SETE, gender REG
Brocato et al. (2015) SETE, gender of teacher and course level REG
Centra (2003) Difficulty level, workload REG
Figlio and Lucas (2004) Grading pattern of the teacher REG
Galbraith et al. (2012) SETE, class size CLA
Goos and Salomons (2016) SETE REG
Khan and Ghosh (2016) SETE ARM
Khan and Ghosh (2018) SETE, number of student evaluation ARM
Lin et al. (2018) Size of the class taught by the teacher REG
Macfadyen et al. (2015) SETE, gender and age of teacher REG
Nikolic et al. (2015) Quality of notes, facilities and equipments REG
in laboratory
Pandey and Pal (2011) Language used for teaching ARM
Pong-Inwong and Rungworawut (2012) Open-end questions of SETE TM, CLA
Price et al. (2017) SETE, gender of the teacher and student REG
Rani and Kumar (2017) Students’ comment on teaching TM
Stronge et al. (2007) Instruction type, SETE, personal qualities REG
Üstünlüoğlu (2016) Nationality and gender of teacher REG
Uttl et al. (2017) SETE REG
Yin et al. (2016) SETE in theoretical courses REG
ARM: Association rule mining, CLA: Classification, REG: Regression, TM: Text mining
The researchers have analysed the impact of students’ behaviour on their perfor-
mance as well. In a study, Papamitsiou et al. (2014) have found the influence of
students’ temporal behaviour while answering the questions during the examination.
Even the spatial position in the laboratory influences student performance (Ivančević
et al. 2010). Campagni et al. (2015) have analysed the performance relating to the
order of examination taken by the student. The result shows that superior students
took most exams according to the order planned by the curriculum. Dvorak and Jia
(2016) find that students can earn higher grades if they start working on assign-
ments earlier. In another interesting study, Pal and Chaurasia (2017) observe the
negative effect of alcohol consumption on student performance. Some studies have
even reported the possible impact of cell phone usage (Felisoni and Godoi 2018),
social behaviour (Bayer et al. 2012; Carter et al. 2017), and active participation in
the online discussion forum (Romero et al. 2013). Several universities nowadays use
learning management tools for enhancing the knowledge dissemination process. The
data extracted from these tools have helped some studies (Romero et al. 2008; Van
Inwegen et al. 2015; Wang et al. 2016) to analyse student performance as well. In a
similar work, Zacharis (2015) finds the frequency of reading and posting messages,
Table 5 Domain knowledge related studies
Adjei et al. (2016) Prerequisite skills REG

Bydžovská (2016) Performance in similar courses CTR
Chen et al. (2017) Performance in prerequisite CTR
Damaševičius (2010) Difficulty level of the course topic ARM
She et al. (2012) Previous performance in related course REG
ARM: Association rule mining, CTR: Clustering, REG: Regression
viewing files, efforts on content creation, solving quiz may be considered as the
predictor of students’ grade. Besides, the learning strategy is also capable of
indicating student performance (Gasevic et al. 2017).
3.1.2 Assessments
EDM researchers have widely explored the cumulative grade point average (CGPA),
internal and external assessments for predicting student performance. The internal
assessment in literature refers to the internally evaluated scores in assignments, quiz,
lab work, class test or other similar components. In contrast, the marks obtained in
other courses are well-known as external assessments among researchers (Shahiri
et al. 2015). Chaturvedi and Ezeife (2013) have analysed the association of assign-
ment marks with final grade and found a positive relationship between these two. In
fact, multiple studies have explored various types of internal assessment, like per-
formance in the quiz, laboratory, mid-term examination etc. Most of them found a
positive influence on final performance (Ahmed et al. 2014; Helal et al. 2018; Jishan
et al. 2015; Kaviyarasi and Balasubramanian 2018; Romero et al. 2008). Besides this,
evidence of the positive impact of attendance (Ahmed et al. 2014; Bucos and Druag-
ulescu 2018; Fernandes et al. 2017; Jishan et al. 2015) and teamwork score (Parack
et al. 2012) also exist in the literature.
In addition to internal assessments, the researchers have also explored the exter-
nal assessments for analysing student performance. In one such study, Buldu and
Üçgün (2010) find that the students who could not succeed in a numerical course
can fail again in further attempts. They even observe a high correlation of failure in
mathematics, physics and chemistry. The performance in first-year mathematics or
computer science can act as a performance indicator of other courses later (Gold-
ing and Donaldson 2006; Li et al. 2013). Al-Barrak and Al-Razgan (2016) propose
a mechanism for identifying the most critical mandatory courses that can help in
predicting final grade. The achievements in previous courses are indeed a generic
indicator of student performance (O’Connell et al. 2018). Importantly, the CGPA
considers all prior performances and consolidates them to a single value. It is there-
fore quite evident that many studies could establish CGPA as a significant predictor
of student performance (Buniyamin et al. 2015; Christian and Ayub 2014; Huang and
Fang 2013; Jishan et al. 2015). Some studies have even found the association with
school-level performance (Kamley et al. 2016; Tair and El-Halees 2012), admission
test score (Guarin et al. 2015), previous academic experience (Superby et al. 2006)
as well.
3.1.3 Teaching
Analysing the effect of teaching is a long-debated topic in literature. A majority of

existing studies analyse the impact of teaching on indirect factors such as students’
motivation, satisfaction etc. Only a few studies exist which explores the direct influ-
ence on student performance. Here we present a meta-analysis and a summary (refer
Table 4) of these studies which analyse various aspects of teaching and their direct or
indirect impact on student performance.
In one of the related study, Stronge et al. (2007) mention that teaching excellence
with efficient knowledge dissemination, student assessment, classroom management,
and superior personal qualities helps the student in achieving their desired goal.
Another study (Pong-Inwong and Rungworawut 2012) shows that prior knowledge
of the class sentiment is helpful for the teacher to improve their instruction process
which could, in turn, impact student performance. The value addition is a mea-
sure which can be useful in determining the learning achievement (Stronge et al.
2007). Some studies (Yin et al. 2016; Khan and Ghosh 2016, 2018) indeed observe
that effective teaching motivates the student to perform better whereas performance
degrades with poor teaching. A regression analysis shows that student satisfaction
is related to the quality of notes, facilities and equipment in the laboratory as well
(Nikolic et al. 2015). In another study, Pandey and Pal (2011) observe a significant
impact of the instruction delivery language on class attendance which influences the
performance as well. Higher grading standard by teachers can also be beneficial to
the students. The magnitude of such benefit depends on student quality (Figlio and
Lucas 2004). The class size also influences the teaching quality which in turn affects
the final grade of students (Lin et al. 2018).
Abrami et al. (2007) have mentioned that the relationship between teaching qual-
ity and learner achievement is an exciting research area, and it needs further detail
investigation. However, determining the proper measure of teaching quality is still
a research challenge (Khan and Ghosh 2018). Importantly, student evaluation of
teaching excellence (SETE) is a powerful instrument which can help to overcome
this challenge. Üstünlüoğlu (2016) indeed put more emphasis in favour of SETE
related research. However, the author reminds the necessity of student, employer and
teacher-centric SETE. In addition to formal measures, sentiment analysis of student
feedback is an indirect assessment which facilitates the teachers to assess the stu-
dents’ interest in class (Rani and Kumar 2017). However, the community is sceptical
about the validity of such evaluation. For example, Uttl et al. (2017) do not find a
strong correlation between learning achievement and teacher rating. Galbraith et al.
(2012) observe the ability of SETE in predicting student learning achievement though
raises a concern about possible bias of class size. Goos and Salomons (2016) also
reminded about the inaccuracy of SETE due to varying response rate across courses.
Not only that, the bias between student gender, age, specialisation area, final grade
and SETE is also evident in the literature (Macfadyen et al. 2015; Price et al. 2017;
Zabaleta 2007).
In contrast, some researchers consider SETE as a reliable, stable and rela-
tively valid indicator of effective teaching (Balam and Shannon 2010; Moore and
Kuol 2005; Khan and Ghosh 2016, 2018). Marsh (2007) thinks that SETE may be
considered as a useful instrument for teachers to gather feedback on teaching, student
to select courses, and administration to make decisions. However, a systematic feed-
back collection process is necessary, and proper analysis of it can help in many ways
(Hattie and Timperley 2007; Jara and Mellar 2010). Although possible gender bias
is evident in the literature, Brocato et al. (2015) report that it is not very significant
in a traditional classroom setting. Centra (2003) even rejects the myth of possible
bias between final grade and SETE. Importantly, Marsh (1984) has explained the rea-
son behind this myth long ago. According to the author, the co-occurrence of higher
ratings and higher grades might be a result of better learning due to more effective
teaching.
3.1.4 Domain knowledge
According to Blooms’ taxonomy (Bloom et al. 1956), knowledge involves the recall
or recognition of terms, ideas, processes, and theories. In the learning context, domain
knowledge is the cognition on a similar topic learned earlier. However, quantify-
ing the domain knowledge is a challenge on its own (Lorenzetti et al. 2016). The
researchers have used the domain knowledge to analyse various areas such as self-
regulated learning (Moos and Azevedo 2008), web-based problem-solving pattern
(She et al. 2012), reading comprehension (McCarthy and Goldman 2017), internet
searching for learning objects (Willoughby et al. 2009) etc. These studies primar-
ily measure the domain knowledge based on previous experience, specialisation, or
through a set of questionnaires. The syllabus is available in digital format nowadays,
which contains valuable information about the course content. The course content
indeed represents the domain knowledge repositories and cognitive models of the
knowledge components to be learned (Chung and Kim 2016; Pen̄a-Ayala 2014).
However, only a limited number of studies measuring the direct impact of domain
knowledge on student performance exists in the literature. Here we present a few
closely related studies and summarises them in Table 5.
In one of the few studies available, Adjei et al. (2016) show that availability of
prerequisite skills affect student performance. Chen et al. (2017) have used a pre-
requisite dataset to measure domain knowledge which facilitates the prediction of
student response in assessments. The difficulty level of course topic also has a signif-
icant effect on performance. Prior knowledge could help in such cases (Damaševičius
2010). Bydžovská (2016) have clustered the courses based on cosine similarity mea-
sures and later on analyses the student performance. They found that these cluster
of courses can help in performance prediction. In another study, She et al. (2012)
use past performance in related courses to quantify the domain knowledge. Even
they choose similar courses manually but able to establish an association of domain
knowledge with student performance.
3.2 Methods used for identifying the factors
Researchers have applied various techniques such as clustering, classification, asso-

ciation rule mining etc. for extracting valuable knowledge from the educational
dataset. This subsection briefly discusses the EDM approaches applied in the existing
literature.
3.2.1 Association rule mining
Agrawal et al. (1993) introduced the concept of association rule mining for market
basket analysis. Later on, it was widely used in various other domains as well. The
association rule mining is one of the well-known and popular data mining techniques
used extensively for educational purposes (Kamley et al. 2016). It is beneficial indeed
for understanding the pedagogical aspects of learning which in turn help the academic
administrator to frame policies (Angeli et al. 2017; Damaševičius 2010; Garcı́a et al.
2007). There is a reasonable number of studies exist which use the association rule
mining for analysing student performance (Ahmed et al. 2014; Buldu and Üçgün
2010; Chaturvedi and Ezeife 2013; Damaševičius 2010; Garcı́a et al. 2007; Kamley
et al. 2016; Pandey and Pal 2011; Parack et al. 2012; Tair and El-Halees 2012).
In order to obtain a set of important rules, it is essential to pre-determine the min-
imal support and confidence. However, it is difficult for an educator to decide these
two input parameters in advance. Furthermore, the number of obtained rules may be
too high in some cases, and most of them are non-interesting and with low compre-
hensibility. Instead of filtering, the ranking of important rules may be one possible
solution to address this problem (Garcı́a et al. 2007). A combined measure of cumula-
tive interestingness may also be effective in this context (Damaševičius 2010). Above
all, a non-parametric technique of data mining can be more useful for the end-users
in educational settings (Zorilla et al. 2010).
3.2.2 Regression
The regression is a prevalent technique for statistical analysis, and it is quite popular
in the data mining context as well. It is probably the most straightforward method for
investigating the functional relationship between variables (Kotsiantis and Pintelas
2005). An equation between the dependent variable and one or multiple independent
variables defines the relationship in regression analysis. The equation is the most crit-
ical product here. In prediction settings, it is also used for evaluating the importance
of individual predictors and understand the correlation among variable (Chatterjee
and Hadi 2015; Lipovetsky and Conklin 2015).
The widespread application of regression is evident in educational data mining lit-
erature. It is indeed frequently used in EDM studies for establishing or disproving the
impact of teaching quality. Many research works have explored the regression anal-
ysis for student performance analysis (Abrami et al. 2007; Adjei et al. 2016; Centra
2003; Goos and Salomons 2016; Li et al. 2013; Macfadyen et al. 2015; Papamitsiou
et al. 2014; Stronge et al. 2007; Üstünlüoğlu 2016; Yin et al. 2016; Zabaleta 2007).
3.2.3 Classification
It is one of the primary data mining technique widely used for predicting group
membership of data instances. Researchers have widely applied this method of clas-
sifying each item in a dataset into one of the predefined class or group (Kesavaraj
and Sukumaran 2013). It is possibly the most widely used data mining techniques
in the educational context as well. To support this, Pen̄a-Ayala (2014) mentions that
42.15 % of EDM research has applied the classification technique between 2010
and the first quarter of 2013. There are several classification approaches available
in literature such as decision tree methods, rule-based classification, memory-based
learning, neural networks, bayesian network, and support vector machines. The
decision tree provides a powerful formalism for representing comprehensible and
accurate classifier (Quinlan 1990). The decision tree classification is the most popular
method used in educational research (Shahiri et al. 2015).
The predictive ability of the classification technique makes it a superior choice
for student performance analysis. The primary objective of classification in EDM
is to identify the critical factors that contribute to the final grade of the student
(Al-Barrak and Al-Razgan 2016). Moreover, student classification can help teach-
ers to understand the specific requirements and determine an appropriate approach
as well (Hidayah et al. 2013). There are various studies available in EDM liter-
ature that uses classification approaches for performance predictor identification
(Al-Barrak and Al-Razgan 2016; Buniyami et al. 2015; Christian and Ayub 2014;
Gray et al. 2014; Jishan et al. 2015; Guarin et al. 2015; Mishra et al. 2014; Natek and
Zwilling 2014; Pong-Inwong and Rungworawut 2012; Quadri and Kalyankar 2010;
Ramesh et al. 2013; Romero et al. 2008, 2013).
3.2.4 Clustering
It is a concept of dividing data into groups of similar objects where each group, or
cluster, consists of objects that are similar to one another and dissimilar to the object
of other groups (Berkhin 2006). It is a data modelling technique which has a historical
milestone in the field of mathematics and statistics. However, in machine learning,
clusters correspond to hidden patterns, the search for a cluster is unsupervised learn-
ing, and the resulting system represents the knowledge (Kaufman and Rousseeuw
2009). Partitioning and hierarchical methods are the two broad categories of clus-
tering, whereas k-means and agglomerative hierarchical techniques are very popular
among them.
Although clustering is popular in data mining literature, the researchers have
reported some limitations of clustering approaches. These are quite relevant for edu-
cational datasets as well. Traditional clustering approaches are not suitable when both
quantitative and qualitative attributes coexist (Chen et al. 2016). However, most of
the educational dataset contains both of these types. Most of the clustering approach
additionally require an input parameter which determines the number of resulting
cluster. It is difficult for an educator to determine this parameter (Bogarin et al. 2014).
Nevertheless, EDM researchers have considerably used this unsupervised technique
in student performance analysis literature (Bydžovská 2016; Campagni et al. 2015;

Helal et al. 2018; Ivančević et al. 2010; Romero et al. 2013; Saxena and Govil 2009).
3.2.5 Text mining and social network analysis
Text mining is the full or partially automated process of extracting hidden information
from a large amount of unstructured data. Although text mining is a part of the gen-
eral data mining, it differs from other approaches. It extracts the patterns from natural
language text rather than from structured data (Delen and Crossland 2008; Feldman
and Sanger 2007; Romero and Ventura 2007). Moreover, the social network is a
web-based service commonly used for information dissemination, personal activities
posting, product reviews, online pictures sharing, professional profiling, advertise-
ments, sentiment expression etc. The social network is getting tremendously popular
in the last decade (Zuber 2014). Nowadays, people are relying more on the social net-
work for interaction with other users. It, in turn, generates massive data characterised
by three computational issues, namely, size, noise and dynamism. Data mining pro-
vides various techniques for extracting useful knowledge from these massive datasets
(Kagdi et al. 2007).
Researchers have identified some topics on text mining and social network anal-
ysis, which are growing popularity nowadays. Community detection, semantics,
opinion and sentiment analysis are a few examples among them (Zuber 2014). Inter-
estingly these are very relevant in education as well. Detecting student communities
and analysing community performance can be a thought-provoking study. Similarly,
analysis of student or teacher sentiment may open a new direction of understanding
the knowledge dissemination process. The answer script, textual feedback, communi-
cation between student and teacher can turn out to be valuable inputs for such analysis
(Pen̄a-Ayala 2014). A few interesting EDM studies are available on text mining and
social network analysis (Akçapinar 2015; Bayer et al. 2012; Chung and Kim 2016;
Foley and Allan 2016; Montuschi et al. 2015; Pong-Inwong and Rungworawut 2012;
Rani and Kumar 2017; Rekha et al. 2012; Romero et al. 2013). However, these are
the initial steps only, and there is ample scope of exploring the educational data with
the help of these technologies further (Yim and Warschauer 2017). The application
of semantic approaches has a huge potential in education as it manages the knowl-
edge objects through unstructured data like syllabus, textbook, question paper etc.
(Pen̄a-Ayala 2014). Although some semantic studies are available in EDM literature
(Elouazizi et al. 2017; Hsiao et al. 2016; Hsiao and Lin 2017; Liu et al. 2018a; Mon-
tuschi et al. 2015; Nakayama 2016), it is incipient in the field of student performance
analysis.
4 Student performance prediction
The final grade is supposed to summarise how well a student understands and apply
the knowledge conveyed in a course. However, predicting the actual grade is a
challenging task. This fact has energised more EDM researchers to discover an effi-
cient model of student performance (Bresfelean et al. 2008; Guo et al. 2015; Meier
et al. 2016; Márquez-Vera et al. 2013). In addition to efficiency, time of predic-

tion is another crucial aspect which facilitates the decision-makers and teachers in
taking a right and timely measures (Quille and Bergin 2018). The performance mod-
elling studies use various attributes as the predictor. However, only a few of these
are available before course commencement which creates more challenges in early
prediction.
4.1 Time of prediction
In this part, we present the existing performance modelling literature classified as per
their temporal nature of prediction. It separately presents the meta-analysis of studies
that tries to predict performance after or before course commencement. Besides this,
Tables 6, 7, 8 and 9 summarises them in tabular format.
4.1.1 During the tenure of the course
This survey observes a significant number of studies which predict the performance
during the tenure of the course efficiently. Tables 6 and 7 provides a list of these
studies with the details on predictor and method used, aim and observed efficiency.
It is quite apparent that efficient predictors are required to yield a better result. For-
tunately, many predictor data are readily available during the tenure of the course.
According to Huang and Fang (2013), the CGPA is the sole attribute capable of pre-
dicting the overall academic performance of a course and the grades in prerequisites,
mid-term examination score and the CGPA are the most suitable predictors for the
individual outcome. A majority of studies have indeed used these as predictors. A
few studies have used only internal assessments for predicting student performance
(Hasheminejad and Sarvmili 2018; Meier et al. 2016; Sivakumar and Selvaraj 2018).
Some studies use external assessment like grades in other courses and prerequisites
along with internal assessment records for a better result (Guo et al. 2015; Jishan et al.
2015; VeeraManickam et al. 2018; Yu et al. 2018). Some studies use data on student
demographics and socio-economical factors along with internal assessment as well
(Kotsiantis and Pintelas 2005; Koutina and Kermanidis 2011; Márquez-Vera et al.
2013; Natek and Zwilling 2014; Pandey and Taruna 2016; Santana et al. 2017; Uddin
and Lee 2017; Xu et al. 2017). Researchers have also used student behavioural data
captured from various online tools in this regard. For example, Hong et al. (2017), Lu
et al. (2018) and Yang et al. (2018) have used video viewing behaviour for estimat-
ing student performance. Student behaviour in an online forum (Mueen et al. 2016;
Ornelas and Ordonez 2017; Romero et al. 2008; Widyahastuti and Tjhin 2017; Yoo
and Kim 2014; Yu et al. 2018), learning management system (Conijn et al. 2017; Kim
et al. 2018; Ostrow et al. 2015; Sandoval et al. 2018; Xing et al. 2015), their move-
ment pattern (Zhang et al. 2018), and activity during web browsing (Chaturvedi and
Ezeife 2017) also help in predicting student performance. Thai-Nghe et al. (2009)
and Zollanvari et al. (2017) have used teaching quality and psychological factors of
students for classifying student performance.
In addition to predictors, researchers have explored various methods for predict-
ing student performance as well. Some of them (Hämäläinen and Vinni 2006; Hong
Table 6 Classification studies during the tenure of the course
Study Predictor used Aim Efficiency
Ahmed and Sadiq (2018) Gender, internal assessment, Pass-fail Acc: 83.56 %
attendance
Bresfelean et al. (2008) Student profile and questionnaire Pass-fail Acc: 76.00 %
Cen et al. (2016) Student interaction Pass-fail Acc: 80.00 %
Chaturvedi and Ezeife (2017) Web usage pattern, internal Pass-Fail Acc: 96.00 %
assessment
Chen et al. (2018) Student interaction data Final grade Acc: 70.00 %
Grivokostopoulou et al. (2014) Internal assessment, gender, Pass-fail Acc: 97.80 %
year of study
Guo et al. (2015) Background, internal and external Final grade Acc: 77.20 %
assessment
Hämäläinen and Vinni (2006) Performance in exercise Pass-fail Acc: 80.00 %
Hasan et al. (2018) Student academic and activity data Category Acc: 100 %
Hasheminejad and Internal assessment Final score Acc: 92.00 %
Sarvmili (2018)
Hong et al. (2017) Video viewing behaviour Good-poor AUC: 0.8100
Jishan et al. (2015) Internal and external assessment Final grade Acc: 75.00 %
Kim et al. (2018) Learning activity data Pass-fail AUC: 96.00 %
Koutina and Kermanidis Student demographics, internal Category Acc: 85.70 %
(2011) assessment
Meier et al. (2016) Internal assessment Good-poor Acc: 76.00 %
Mueen et al. (2016) Internal assessment and forum Pass-Fail Acc: 86.00 %
participation
Natek and Zwilling (2014) Internal assessment, demographics, Final grade Acc: 90.00 %
extra-curricular activities
Ornelas and Ordonez (2017) Student interaction data Pass-fail Acc: 90.00 %
Pandey and Taruna (2016) Student demographics and Pass-fail Acc: 87.84 %
assessment data
Quille and Bergin (2018) Student background and Pass-fail Acc: 71.00 %
psychological data
Romero et al. (2008) Time spend in quiz, assignment Final grade Acc: 67.02 %
Santana et al. (2017) Background, assessments Pass-fail FM: 0.8300
Sivakumar and Selvaraj (2018) Internal assessment Category Acc: 98.56 %
Thai-Nghe et al. (2009) Teaching quality Pass-fail FM: 0.6150
Uddin and Lee (2017) Background, external assessment, Pass-fail Acc: 70.00 %
social network interaction
Xu et al. (2017) Background, internal assessment Rank Acc: 25.00 %
Yoo and Kim (2014) Participation in online forum Final grade Acc: 91.40 %
Yu et al. (2018) Usage pattern of online forum, internal Pass-fail Acc: 78.08 %
and external assessment, attendance
Table 6 (continued)
Xing et al. (2015) Log data from geometry learning tool Pass-fail Acc: 80.20 %
Zhang et al. (2018) Student movement pattern Good-poor Acc: 95.00 %
Zollanvari et al. (2017) Psychological factors High-low Acc: 82.00 %
Acc: Accuracy, AUC: Area under curve, FM: F-measure
et al. 2017; Huang and Fang 2013) observe that the applied method does not influ-
ence the prediction efficiency. In contrast, many studies have compared the efficiency
of various algorithms and reported the superiority of a few among them. Quite a few
studies have observed better result with decision tree classifier (Cen et al. 2016; Gri-
vokostopoulou et al. 2014; Natek and Zwilling 2014; Chaturvedi and Ezeife 2017;
Romero et al. 2008). The superiority of naı̈ve bayes (Jishan et al. 2015; Ornelas and
Ordonez 2017), artificial neural network (Jishan et al. 2015; Widyahastuti and Tjhin
Table 7 Regression studies during the tenure of the course
Cen et al. (2016) Student interaction Final grade MAE: 0.0800

Conijn et al. (2017) Internal assessment and usage Final grade R 2 : 0.4300
pattern of moodle
Huang and Fang (2013) CGPA, internal and external Final score APA: 90.10 %
assessment
Kotsiantis and Pintelas (2005) Student demographics, internal Final-score MAE: 1.21
assessment
Lu et al. (2018) Internal assessment, video Final score R 2 : 0.4000
viewing behaviour
Márquez-Vera et al. (2013) Background, external assessments, Pass-fail Acc: 98.70 %
socio-economic factors
Ostrow et al. (2015) ASSISTment usage pattern Final score R 2 : 0.9610
Sandoval et al. (2018) Background, academic records, Final score MAE: 0.0615
behaviour in learning manage-
ment system
VeeraManickam et al. (2018) Internal and external assessment Final score RMSE: 4.665
Widyahastuti and Tjhin (2017) Attendance, usage pattern of Final grade MAE: 0.1185
online forum
Yang et al. (2018) Internal assessment, video Final score Acc: 80.00 %
viewing behaviour
Acc: Accuracy, APA: Average prediction accuracy, MAE: Mean absolute error, RMSE: Root mean square
error
Table 8 Classification studies before course commencement
Chanlekha and Niramitranon Student demographics, performance At-risk Acc: 62.50 %

(2018) in school and admission
Garcia and Mora (2011) Student demographics, social life, Category Acc: 58.64 %
and result in diagnostic test
Hidayah et al. (2013) Intelligence level, interests, talents, Category RMSE: 0.2561
motivation
Kabakchieva (2012) Social and academic background, Category Acc: 73.59 %
previous failures
Kabra and Bichkar (2011) High school performance, family Pass-fail Acc: 69.94 %
background
Martinez (2001) High school performance, student Pass-fail Acc: 62.60 %
background
Mimis et al. (2018) Gender, motivation, previous Final grade Acc: 52.32 %
performance
Mishra et al. (2014) Social, academic background, Pass-fail Acc: 94.42 %
emotional skills
Osmanbegović and Suljić Social and academic background, Pass-fail Acc: 76.65 %
(2012) attitudes towards study
Ramesh et al. (2013) Gender, social and academic Final grade Acc: 72.00 %
background
Sullivan et al. (2017) Socio-economic factors Below-above Acc: 86.00 %
Acc: Accuracy, RMSE: Root mean square error
2017; Yu et al. 2018), support vector machine (Santana et al. 2017), random for-
est (Ahmed and Sadiq 2018; Chen et al. 2018; Hasan et al. 2018; Sandoval et al.
2018), deep learning (Guo et al. 2015; Kim et al. 2018) and linear regression (Lu
et al. 2018; Yang et al. 2018) are also evident in the literature. Apart from the appli-
cation of traditional data mining approaches, researchers have also proposed some
specialised algorithms to predict student performance (Hasheminejad and Sarvmili
2018; Márquez-Vera et al. 2013; Meier et al. 2016; Uddin and Lee 2017; Xu et al.
2017; Zollanvari et al. 2017). Some of them have evolved from state-of-the-art
algorithms after customisation, whereas a few are entirely new approach.
Moreover, these studies have reported significant efficiency. Some of them even
achieved more than 90 % accuracy in classification settings (Chaturvedi and Ezeife
2017; Grivokostopoulou et al. 2014; Hasan et al. 2018; Hasheminejad and Sarvmili
2018; Márquez-Vera et al. 2013; Natek and Zwilling 2014; Ornelas and Ordonez
2017; Sivakumar and Selvaraj 2018; Yoo and Kim 2014; Zhang et al. 2018). Some
of these studies report high efficiency, but that may be due to the imbalanced dataset
used. Unfortunately, many of the studies available in the literature ignore the class
Table 9 Regression studies before course commencement
Backenköhler and Wolf (2017) CGPA, age, semester, number Final grade MSE: 0.4106
of attempt
Bahritidinov and Sánchez (2017) Domain knowledge, workload, Final grade RMSE: 0.7255
availability
Cakmak (2017) Grades in similar courses Final grade MAE: 0.2600
Elbadrawy et al. (2016) Student demographics, content and Final score RMSE: 0.7443
grades of previous courses
Gray et al. (2014) Psychometric factors Pass-fail Acc: 73.30 %
Ibrahim and Rusli (2007) Previous knowledge, performance, Final score RASE: 0.1714
background
Pardos et al. (2010) Skills identified through Final score MAD: 4.210
coding test
Polyzou and Karypis (2016) Grades in last semester Final grade RMSE: 0.6710
Sweeney et al. (2016) Grades in previous semester Final grade RMSE: 0.7758
Acc: Accuracy, MAE: Mean absolute error, MSE: Mean square error, RASE: Root average squared error,
RMSE: Root mean square error, MAD: Mean average deviance
imbalance issue. Although Hasan et al. (2018) have predicted all instances correctly,
it is essential to mention here that they have used records of a few students only.
4.1.2 Before course commencement
Although the EDM literature on performance prediction during the tenure of the
course is mature enough, prediction before course commencement is still incipient.
Here we present a meta-analysis of these studies, where Tables 8 and 9 provide a list
of them with relevant details.
The different characteristics of predictors make the early prediction studies differ-
ent from the former type of studies. In the present scenario, the predictors should be
available before course commencement. Collecting the relevant data and their mea-
surement is the major challenge here. As an example, Hidayah et al. (2013) tried
to predict student performance based on their level of intelligence, interests, talents,
and motivation. It collects the data using a questionnaire and quiz at the beginning
of the semester. Gray et al. (2014) have used psychometric indicators like person-
ality, motivation and learning strategies in a similar study. It is not feasible always
to collect data through a questionnaire as it requires students’ involvement. There-
fore, EDM researchers have widely used the student demographics data and past
performance in many early prediction studies (Backenköhler and Wolf 2017; Cak-
mak 2017; Chanlekha and Niramitranon 2018; Elbadrawy et al. 2016; Garcia and
Mora 2011; Ibrahim and Rusli 2007; Kabakchieva 2012; Kabra and Bichkar 2011;
Martinez 2001; Mimis et al. 2018; Polyzou and Karypis 2016; Sweeney et al. 2016).
Some studies have even utilised the domain knowledge for this purpose (Bahritidi-
nov and Sánchez 2017; Ibrahim and Rusli 2007). Besides this, researchers have used
socio-economic factors (Sullivan et al. 2017), prior skills (Pardos et al. 2010), and
performance in the diagnostic test (Garcia and Mora 2011) as well.
Tables 8 and 9 present the multiple prediction studies which have applied classifi-
cation and regression approach respectively. In one of these studies, Osmanbegović
and Suljić (2012) observe that naı̈ve bayes, artificial neural network, and decision tree
classifier perform almost equally well for classification. In contrast, some of them
have reported superiority of artificial neural network (Kabakchieva 2012; Mimis et al.
2018; Ramesh et al. 2013), support vector machine (Gray et al. 2014) and random
forest (Mishra et al. 2014) algorithms. Besides this, the usage of advanced techniques
like fuzzy neural network (Hidayah et al. 2013), matrix factorization (Sweeney et al.
2016), collaborative filtering (Cakmak 2017), and decision tree with entropy split-
ting (Sullivan et al. 2017) are also evident in the literature. Importantly, the efficiency
of these studies is not as high as observed in prediction studies during the tenure of
the course. In the case of classification, only Mishra et al. (2014) and Sullivan et al.
(2017) have reported more than 80 % accuracy. However, Mishra et al. (2014) have
used a dataset which contains only 215 records of single specialisation for analysis.
Sullivan et al. (2017) have used socio-economic factors like free and reduced lunch
receiving percentage, jobless rate, crime rate etc. However, the data for some of these
factors may not be available in all circumstances.
4.2 Aim of prediction
In this section, we outline the aim or broad objective of the prediction studies dis-
cussed earlier. These studies have tried to achieve various objectives which can be
grouped broadly in two categories. Some of them try to predict the student perfor-
mance either in binary classes like pass-fail, success-failure etc., whereas others have
attempted to predict the performance in terms of the final grade or actual score.
Prediction in binary terms seems to be more popular amongst these two categories.
4.2.1 Success prediction
Modelling the actual grade or marks is a challenging task as it depends on diverse

factors such as demographics, educational background, personal, psychological, aca-
demic progress and other environmental variables (Guo et al. 2015). Relationship
between many of these variables are yet to understand and, therefore, performance
classification turns out to be more popular among researchers. It is quite reason-
able that a prediction is more straightforward with fewer target classes (Hu et al.
2017). Many researchers have tried to classify the student performance in binary
terms such as pass-fail, below and above a reference level, good-poor etc. (Ahmed
and Sadiq 2018; Cen et al. 2016; Chaturvedi and Ezeife 2017; Gray et al. 2014; Hong
et al. 2017; Meier et al. 2016; Sullivan et al. 2017; Zollanvari et al. 2017). However,
some studies try to classify in more generic terms with more than two class values
(Hasan et al. 2018; Koutina and Kermanidis 2011; Sivakumar and Selvaraj 2018;
Garcia and Mora 2011). Importantly, Gardner and Brooks (2018) mention that non-
parametric tree-based algorithms can be more effective in success prediction rather
than techniques which require hyper-parameter tuning. Even the ensemble meth-
ods can provide a promising result as their predictive performance is generally high
(Miguéis et al. 2018).
4.2.2 Grade or marks prediction
As mentioned earlier, some researchers have even tried to predict actual marks (Guo
et al. 2015; Huang and Fang 2013; Lu et al. 2018; Romero et al. 2008; Sandoval et al.
2018; Yang et al. 2018). The existing studies have used both regression and classifi-
cation techniques for this purpose. A majority of them use the internal assessment as
input data to predict grade or marks during the tenure of the course. A few prediction
studies before course commencement tried to estimate the final grade or marks as
well (Backenköhler and Wolf 2017; Bahritidinov and Sánchez 2017; Cakmak 2017;
Elbadrawy et al. 2016; Mimis et al. 2018; Polyzou and Karypis 2016; Ramesh et al.
2013; Sweeney et al. 2016). However, they are not as efficient as the previous one.
Moreover, Hu et al. (2017) observe that the number of studies which attempt to
predict actual grade or score is comparatively less than the success prediction cases.
5 Discussion
In this paper, we have cited 207 studies out of which 192 are published by EDM
researchers either for analysing student performance or they are related to EDM
techniques or applications. We have discussed the previous 16 surveys which are
related to this topic and reviewed 140 core studies on student performance analysis
or prediction in classroom-based education. The primary focus in 79 of them is to
find performance influencing factors and 61 studies try to predict the student per-
formance. Figure 3 presents the year wise count of these works. It shows that data
mining studies on student performance have gained significant momentum in recent
years, especially since 2010. Figure 4 additionally presents the percentage of various
factors and methods used by 79 studies on student performance analysis. It seems
that student background, behaviour and their internal or external assessments are
most popular, whereas the researchers put the least focus on their domain knowl-
edge. Regression and classification are found to be the most popular method used in
these student performance related studies. Besides this, Fig. 5 shows the distribution
of 61 performance prediction studies based on their objective and time of prediction.
We find only 33 % of these studies capable of predicting student performance before
course commencement. Furthermore, success or failure prediction turns out to be
more popular than the prediction of final grade or score.
As mentioned earlier, predictor identification is one of the most critical tasks in
any modelling study, and it is quite prevalent in student performance prediction as
well (Helal et al. 2018). A handful of the existing literature explores the student-
related factors such as their quality, how they perform in assignment and class test,
whether they attend classes, what was their social and academic background etc. We
Fig. 3 Year-wise number of studies
also found some conflicting observation about the influence of gender on student per-
formance. However, we could not find any study which reports a significant influence
of gender during prediction. There is no doubt that the quality of a student and their
background creates quite a significant impact on student performance. It could be one
of the possible reasons that the research community is extremely focused on student-
related factors. However, there are many other external factors which could have
a significant influence on performance. The teaching quality and previous domain
knowledge are two among them. A few existing studies have found a positive impact
Fig. 4 Percentage of various factors and methods used in student performance analysis
Fig. 5 Percentage of performance prediction studies as per their aim and time of prediction
of teaching on student motivation. Some of them have even established a strong asso-
ciation between teaching quality and student performance. However, we have found
only one prediction study which has used the teaching quality as a predictor. The
domain knowledge of a student is also less utilised while predicting student perfor-
mance. It is essential to mention here that the teaching quality and domain knowledge
are difficult to measure. Although student evaluation of teaching excellence (SETE)
can be a possible way to measure teaching quality, the research community is scep-
tical about the validity of the same. The possible bias between student performance
and SETE is the primary reason behind this. It is worthy to point out that Khan and
Ghosh (2018) have already found a strong association of student performance and
SETE where the authors have taken utmost care to reduce the bias. Besides this,
researchers have widely used questionnaire, observational techniques etc. for mea-
suring domain knowledge. However, the data collected through these methods have
some limitations (Conijn et al. 2017). The readily available course syllabus data can
help in this regard.
Importantly, the performance modelling studies have utilised various factors as
predictor, and the list of these predictors is not a short one (refer Tables 6 – 9).
However, the question is which of them are the candidate of best predictor set. It
is still an open research question. In order to estimate an answer, we have com-
pared the effectiveness of various broad predictor types and presented the findings in
Table 10. Significantly, the performance prediction studies evaluate their efficiency
using various metrics. The prediction accuracy is most popular among them. There
are, in fact, 40 out of 61 studies which have used prediction accuracy to measure the
model efficiency. We, therefore, consider only these 40 studies for comparison. The
findings show that student behaviour measured during the tenure of the course and
internal assessment together can predict performance with an average accuracy of
88.02 %. The separate use of student behaviour and internal assessment is even capa-
ble of yielding more than 80 % average accuracy. The marks in the assignment, quiz,
class test etc., attendance, learning behaviour, activities in various educational sup-
port tools are a few examples of such predictors. Although they turn out to be very
useful during performance prediction, they can not help in early prediction before
course commencement. As per the findings reported in Table 10, student background
Table 10 Comparison of various predictors used in student performance prediction studies
No of studies * Avg. Accuracy
Predictor type After Before After Before
Student background 0 1 – 86.00%

Student behaviour 8 1 81.95% 73.30%
Assessment 5 0 84.31% –
Student background, behaviour 2 1 73.50% 94.42%
Student background, assessment 8 5 80.73% 68.13%
Student behaviour, assessment 5 0 88.02% –
Student background, behaviour, assessment 1 3 70.00% 62.54%
Overall 29 11 82.07% 71.09%
*40 out of 61 performance prediction studies measure efficiency in terms of accuracy
and behaviour like emotional skills are most effective while predicting student per-
formance before course commencement. However, only one such study exists which
has used these predictors together. The researchers have used external assessments in
early prediction studies as well, but they are not as effective as internal assessment
marks. The overall average accuracy observed in early prediction is only 71.09 %
as compared to 82.07 % for studies predicting performance during the tenure of the
course. More comprehensive research is therefore needed to search for more influ-
encing factors which can be estimated early. The semi-structured or unstructured
data generated by social media, IoT (internet of things) devices, student movement
pattern, and logs of several educational support software may help in this regard.
Although existing studies have widely used classification and regression techniques
on the structured data, text mining on semi-structured or unstructured data can be the
future research direction.
Furthermore, a majority of 46 % modelling studies prefer to classify the perfor-
mance as success or failure. The efficiency of such classification is usually high due
to a fewer number of class labels. A few of them even achieved more than 90 % pre-
diction accuracy. Even so, many of these studies ignore the important concern of class
imbalance. If there are only 5 % failure cases in an imbalanced dataset, for example,
and none of the predictions is failure, then also the accuracy would be 95 %. Classi-
fication in terms of three or four categories is also present in literature. However, the
number of such studies are very less. Some modelling studies try to classify the final
grade or score as well. Although these studies report significant prediction efficiency
during the tenure of the course, the reported efficiency before course commencement
is comparatively lower. The final score prediction before course commencement also
seems to be a research challenge. Undoubtedly, it is a complex task, as various influ-
encing factors are not yet known. Still, students can get the benefit of such early
prediction by taking advantage of the performance-oriented course recommendation
(Bodily et al. 2018). It can additionally help the educators to identify at-risk students
and take a proactive approach in mitigating the risk of retention (Gašević et al. 2016).
Finally, we also like to point out two crucial aspects observed during this survey.
The first one is the widespread use of grade or score for finding performance influ-
encing factors. The examination score is a widely accepted performance indicator
which measures the knowledge in general (Pandey and Pal 2011). Many EDM stud-
ies, therefore, try to find influencing factors by establishing a correlation with the
final grade. However, a superior student is supposed to perform better than a rela-
tively inferior one. The final grade may not be sufficient to analyse the value addition
due to external influencing factors. Khan and Ghosh (2018) have proposed a mecha-
nism to measure the value addition, which can help to find the association of several
external influencing factors. Secondly, the existing literature primarily focuses on
improving the prediction accuracy but ignores the wrongly classified instances. For
example, false-negative cases in failure prediction may lead to various problems. In
such cases, the teacher may overlook a student who needs extra care but predicted to
be successful. The prediction confidence may help in this regard. A statement like ‘I
am 90 % confident that the student will fail’ may be more effective than a statement
like ‘I think the student will fail’.
6 Conclusion
In this paper, we have presented a review of existing EDM literature on student

performance analysis and prediction. It primarily focuses on studies related to
classroom-based education. As an initial step, we have first identified the current and
significant research question in this domain and additionally developed a taxonomy
of the research directions as well. This systematic review has utilised the features of
Mendeley, a reference management tool provided by Elsevier, for categorising and
organising the relevant articles. We have adopted a hybrid approach for searching
relevant articles. In addition to the conventional searching mechanism, the alert sys-
tem of Google Scholar helps in this regard. This review presents a meta-analysis of
performance influencing factors identified by the researchers. It also discusses the
existing student performance prediction studies in the context of their aim and time of
prediction. It seems that the literature on student performance prediction after course
commencement is rich. However, the early prediction before course commencement
is still an open challenge. Table 11 summarises the critical observations of this survey
and also presents the recommendation.
This systematic survey would undoubtedly help the EDM researchers in advancing
the field of next-term grade prediction. Such early prediction can facilitate the design
of various intelligent computer-based educational systems as well. The performance-
aware course recommendation system is one of such example among them. However,
our survey includes research papers published in English only. Studies in other lan-
guages are not considered here due to our lack of proficiency in those languages. It
may be possible that some significant work of student performance prediction before
course commencement is available in other languages, but not mentioned in this sur-
vey. Moreover, we have used Google Scholar for searching the relevant studies. Any
article not listed with Google Scholar is not a part of this survey.
Table 11 Observations and recommendation for future research directions
Observations and recommendations
What are the factors influencing student performance in classroom learning?

Observation
– Association of final performance with student quality, their behaviour, family background,
performance in earlier attended courses, quiz, midterm examination etc. are already established.
– It seems that teaching quality and domain knowledge also influences student performance.
Recommendation
– Future research can focus on measuring other influencing factors before course commencement and
finding their impact on student performance.
– A superior student is expected to get better marks. Therefore, proper measure of value addition, in
place of grade or marks, can help in establishing the impact of other factors.
Which methods are used for finding these factors?
Observation
– Regression and classification are the two methods widely used for establishing the impact of
various factors.
– The usage of semi-structured or unstructured data is comparatively less for this purpose.
Recommendation
– The text mining of unstructured educational data in course syllabus, examination question paper,
answer script etc. can further help in understanding various aspects of learning.
– The social network analysis can help in understanding the student behaviour and their impact
on performance.
Is it possible to predict before course commencement?
Observation
– Some studies have tried to predict performance before course commencement. However, they are less
efficient compared to the prediction studies during the tenure of the course.
– Internal assessment and behaviour seems to be most effective predictor of student performance
during the tenure of the course. Unfortunately, these predictors are not available before course
commencement.
Recommendation
– Future research may focus on student performance prediction before course commencement.
– Researchers should try to improve early prediction efficiency as well. In addition to student
background and past performance, previous teaching quality of the teacher, domain knowledge
and recent behaviour of the student can also be explored for this purpose.
Can we predict the actual grade or score?
Observation
– Researchers have primarily focused on classifying student success or grade. Success prediction is
more popular and efficient among them.
– Only a few studies are available that tries to predict the final score.
Recommendation
– Future studies can put more focus on final grade or score prediction. Improving the success prediction
efficiency before course commencement can be another future research direction.
References
Abrami, P.C., D’Apollonia, S., Rosenfield, S. (2007). The dimensionality of student ratings of instruction:
what we know and what we do not. In The scholarship of teaching and learning in higher education:
an evidence-based perspective (pp. 385–456): Springer.
Adjei, S.A., Botelho, A.F., Heffernan, N.T. (2016). Predicting student performance on post-requisite skills
using prerequisite skill data: an alternative method for refining prerequisite skill structures. In Pro-
ceedings of the sixth international conference on learning analytics & knowledge (pp. 469–473):
ACM.
Aghabozorgi, S., Mahroeian, H., Dutt, A., Wah, T.Y., Herawan, T. (2014). An approachable analytical
study on big educational data mining. In International conference on computational science and its
applications (pp. 721–737): Springer.
Agrawal, R., Imieliṅski, T., Swami, A. (1993). Mining association rules between sets of items in large
databases. ACM SIGMOD Record, 22(2), 207–216.
Ahmed, N.S., & Sadiq, M.H. (2018). Clarify of the random forest algorithm in an educational field. In
2018 international conference on advanced science and engineering (ICOASE) (pp. 179–184): IEEE.
Ahmed, S., Paul, R., Hoque, M.L., Sayed, A. (2014). Knowledge discovery from academic data using asso-
ciation rule mining. In 2014 17th international conference on computer and information technology
(ICCIT) (pp. 314–319): IEEE.
Akçapinar, G. (2015). How automated feedback through text mining changes plagiaristic behavior in
online assignments. Computers & Education, 87, 123–130.
Al-Barrak, M.A., & Al-Razgan, M. (2016). Predicting students final GPA using decision trees: a case
study. International Journal of Information and Education Technology, 6(7), 528–533.
Al-Obeidat, F., Tubaishat, A., Dillon, A., Shah, B. (2017). Analyzing students’ performance using multi-
criteria classification. Cluster Computing, 21(1), 623–632.
Angeli, C., Howard, S., Ma, J., Yang, J., Kirschner, P.A. (2017). Data mining in educational technology
classroom research: can it make a contribution? Computers & Education, 113, 226–242.
Asif, R., Merceron, A., Ali, S.A., Haider, N.G. (2017). Analyzing undergraduate students’ performance
using educational data mining. Computers & Education, 113, 177–194.
Backenköhler, M., & Wolf, V. (2017). Student performance prediction and optimal course selection: an
MDP approach. In International conference on software engineering and formal methods (pp. 40–47):
Springer.
Bahritidinov, B., & Sánchez, E. (2017). Probabilistic classifiers and statistical dependency: the case for
grade prediction. In International work-conference on the interplay between natural and artificial
computation (pp. 394–403): Springer.
Baker, R.S. (2014). Educational data mining: an advance for intelligent systems in education. IEEE
Intelligent Systems, 29(3), 78–82.
Baker, R.S., & Inventado, P.S. (2014). Educational data mining and learning analytics. In Learning
Analytics (pp. 61–75): Springer.
Baker, R.S.J.D., & Yacef, K. (2009). The state of educational data mining in 2009 : a review and future
visions. Journal of Educational Data Mining, 1(1), 3–16.
Bakhshinategh, B., Zaiane, O.R., ElAtia, S., Ipperciel, D. (2017). Educational data mining applications
and tasks: a survey of the last 10 years. Education and Information Technologies, 23(1), 537–553.
Balam, E.M., & Shannon, D.M. (2010). Student ratings of college teaching: a comparison of faculty and
their students. Assessment & Evaluation in Higher Education, 35(2), 209–221.
Bayer, J., Bydzovská, H., Géryk, J., Obsivac, T., Popelinsky, L. (2012). Predicting drop-out from social
behaviour of students. In International conference on educational data mining (EDM).
Beemer, J., Spoon, K., He, L., Fan, J., Levine, R.A. (2018). Ensemble learning for estimating individu-
alized treatment effects in student success studies. International Journal of Artificial Intelligence in
Education, 28(3), 315–335.
Bendikson, L., Hattie, J., Robinson, V. (2011). Identifying the comparative academic performance of
secondary schools. Journal of Educational Administration, 49(4), 433–449.
Berkhin, P. (2006). A survey of clustering data mining techniques. In Grouping multidimensional data
(pp. 25–71): Springer.
Bloom, B.S., Englehard, M., Furst, E., Hill, W., Krathwohl, D. (1956). Taxonomy of educational
objectives: the classification of educational goals. Handbook I: Cognitive Domain.
Bodily, R., Ikahihifo, T.K., Mackley, B., Graham, C.R. (2018). The design, development, and implemen-
tation of student-facing learning analytics dashboards. Journal of Computing in Higher Education,
30(3), 572–598.
Bogarin, A., Romero, C., Cerezo, R., Sánchez-Santillan, M. (2014). Clustering for improving educational
process mining. In Proceedings of the fourth international conference on learning analytics and
knowledge (pp. 11–15): ACM.
Bresfelean, V.P., Bresfelean, M., Ghisoiu, N., Comes, C.A. (2008). Determining students’ academic failure
profile founded on data mining methods. In ITI 2008 - 30th international conference on information
technology interfaces (pp. 317–322): IEEE.
Brocato, B.R., Bonanno, A., Ulbig, S. (2015). Student perceptions and instructional evaluations: a
multivariate analysis of online and face-to-face classroom settings. Education and Information
Technologies, 20(1), 37–55.
Bucos, M., & Druagulescu, B. (2018). Predicting student success using data generated in traditional
educational environments. TEM Journal, 7(3), 617–625.
Buldu, A., & Üçgün, K. (2010). Data mining application on students’ data. Procedia-Social and
Behavioral Sciences, 2(2), 5251–5259.
Buniyamin, N., Mat, U.B., Arshad, P.M. (2015). Educational data mining for prediction and classification
of engineering students achievement. In IEEE 7th international conference on engineering education
ICEED 2015 (pp. 49–53).
Bydžovská, H. (2016). A comparative analysis of techniques for predicting student performance. In
Proceedings of the 9th international conference on educational data mining.
Cakmak, A. (2017). Predicting student success in courses via collaborative filtering. International Journal
of Intelligent Systems and Applications in Engineering, 5(1), 10–17.
Campagni, R., Merlini, D., Sprugnoli, R., Verri, M.C. (2015). Data mining models for student careers.
Expert Systems with Applications, 42(13), 5508–5521.
Carter, A.S., Hundhausen, C.D., Adesope, O. (2017). Blending measures of programming and social
behavior into predictive models of student achievement in early computing courses. ACM Transac-
tions on Computing Education, 17(3), 12.
Cen, L., Ruta, D., Powell, L., Hirsch, B., Ng, J. (2016). Quantitative approach to collaborative learn-
ing: Performance prediction, individual assessment, and group composition. International Journal of
Computer-Supported Collaborative Learning, 11(2), 187–225.
Centra, J.A. (2003). Will teachers receive higher student evaluations by giving higher grades and less
course work? Research in Higher Education, 44(5), 495–518.
Chanlekha, H., & Niramitranon, J. (2018). Student performance prediction model for early-identification
of at-risk students in traditional classroom settings. In Proceedings of the 10th international
conference on management of digital ecosystems - MEDES ’18 (pp. 239–245): ACM.
Chatterjee, S., & Hadi, A.S. (2015). Regression analysis by example. New York: Wiley.
Chaturvedi, R., & Ezeife, C. (2013). Mining the impact of course assignments on student performance. In
Educational data mining 2013.
Chaturvedi, R., & Ezeife, C.I. (2017). Predicting student performance in an ITS using task-driven features.
In 2017 IEEE international conference on computer and information technology (CIT) (pp. 168–175):
IEEE.
Chen, L., Wang, S., Wang, K., Zhu, J. (2016). Soft subspace clustering of categorical data with
probabilistic distance. Pattern Recognition, 51, 322–332.
Chen, W., Brinton, C.G., Cao, D., Mason-singh, A., Lu, C., Chiang, M. (2018). Early detection prediction
of learning outcomes in online short-courses via learning behaviors. IEEE Transactions on Learning
Chen, Y., Liu, Q., Huang, Z., Wu, L., Chen, E., Wu, R., et al. (2017). Tracking knowledge proficiency of
students with educational priors. In Conference on information and knowledge management (CIKM)
(pp. 989–998).
Christian, T.M., & Ayub, M. (2014). Exploration of classification using NBTree for predicting students’
performance. In 2014 international conference on data and software engineering (ICODSE) (pp. 1–6):
IEEE.
Chrysafiadi, K., & Virvou, M. (2013). Student modeling approaches: a literature review for the last decade.
Expert Systems with Applications, 40(11), 4715–4729.
Chung, H., & Kim, J. (2016). An ontological approach for semantic modeling of curriculum and syllabus
in higher education. International Journal of Information and Education Technology, 6(5), 365.
Conijn, R., Snijders, C., Kleingeld, A., Matzat, U. (2017). Predicting student performance from LMS
data: a comparison of 17 blended courses using Moodle LMS. IEEE Transactions on Learning
Cooper, H.M. (1988). Organizing knowledge syntheses: a taxonomy of literature reviews. Knowledge,
Technology & Policy, 1(1), 104–126.
Damaševičius, R. (2010). Analysis of academic results for informatics course improvement using
association rule mining. In Information systems development (pp. 357–363). Berlin: Springer.
Daud, A., Aljohani, N.R., Abbasi, R.A., Lytras, M.D., Abbas, F., Alowibdi, J.S. (2017). Predicting student
performance using advanced learning analytics. In Proceedings of the 26th international conference
on world wide web companion (pp. 415–421): International World Wide Web Conferences Steering
Committee.
Delen, D., & Crossland, M.D. (2008). Seeding the survey and analysis of research literature with text
mining. Expert Systems with Applications, 34(3), 1707–1720.
Dutt, A., Ismail, M.A., Herawan, T. (2017). A systematic review on educational data mining. IEEE Access,
5, 15991–16005.
Dvorak, T., & Jia, M. (2016). Do the timeliness, regularity, and intensity of online work habits predict
academic performance? Journal of Learning Analytics, 3(3), 318–330.
Elbadrawy, A., Polyzou, A., Ren, Z., Sweeney, M., Karypis, G., Rangwala, H. (2016). Predicting student
performance using personalized analytics. Computer, 49(4), 61–69.
Elouazizi, N., Birol, G., Jandciu, E., Öberg, G., Welsh, A., Han, A., et al. (2017). Automated analysis of
aspects of written argumentation. In Proceedings of the seventh international learning analytics and
knowledge conference on - lak ’17 (pp. 606–607): ACM.
Fausett, L.V., & Elwasif, W. (1994). Predicting performance from test scores using backpropagation and
counterpropagation. In IEEE international conference on neural networks, (Vol. 5 pp. 3398–3402):
IEEE.
Feldman, R., & Sanger, J. (2007). The text mining handbook: advanced approaches in analyzing
unstructured data. Cambridge: Cambridge University Press.
Felisoni, D.D., & Godoi, A.S. (2018). Cell phone usage and academic performance: An experiment.
Computers & Education, 117, 175–187.
Ferguson, R. (2012). Learning analytics: drivers, developments and challenges. International Journal of
Technology Enhanced Learning, 4(5/6), 304–317.
Fernandes, E., Carvalho, R., Holanda, M., Van Erven, G. (2017). Educational data mining: discovery
standards of academic performance by students in public high schools in the Federal District of Brazil.
In World conference on information systems and technologies (pp. 287–296).
Fernandes, E., Holanda, M., Victorino, M., Borges, V., Carvalho, R., Erven, G.V. (2018). Educational data
mining: predictive analysis of academic performance of public school students in the capital of Brazil.
Journal of Business Research, 94, 335–343.
Figlio, D.N., & Lucas, M.E. (2004). Do high grading standards affect student performance? Journal of
Public Economics, 88(9-10), 1815–1834.
Foley, J., & Allan, J. (2016). Retrieving hierarchical syllabus items for exam question analysis. In
European conference on information retrieval (pp. 575–586). Cham: Springer.
Galbraith, C.S., Merrill, G.B., Kline, D.M. (2012). Are student evaluations of teaching effectiveness valid
for measuring student learning outcomes in business related classes? A neural network and Bayesian
analyses. Research in Higher Education, 53(3), 353–374.
Garcı́a, E., Romero, C., Ventura, S., Calders, T. (2007). Drawbacks and solutions of applying associa-
tion rule mining in learning management systems. In Proceedings of the international workshop on
applying data mining in e-learning (ADML 2007), Crete, Greece (pp. 13–22).
Garcia, E.P.I., & Mora, P.M. (2011). Model prediction of academic performance for first year students. In
2011 10th Mexican international conference on artificial intelligence (pp. 169–174): IEEE.
Gardner, J., & Brooks, C. (2018). Evaluating predictive models of student success: closing the method-
ological gap. Journal of Learning Analytics, 5(2), 105–125.
Gašević, D., Dawson, S., Rogers, T., Gasevic, D. (2016). Learning analytics should not promote one size
fits all: the effects of instructional conditions in predicting academic success. The Internet and Higher
Education, 28, 68–84.
Gasevic, D., Jovanovic, J., Pardo, A., Dawson, S. (2017). Detecting learning strategies with analytics:
links with self-reported measures and academic performance. Journal of Learning Analytics, 4(2),
113–128.
Gedeon, T.D., & Turner, S. (1993). Explaining student grades predicted by a neural network. In Pro-
ceedings of 1993 international joint conference on neural networks, 1993. IJCNN’93-Nagoya, (Vol. 1
pp. 609–612): IEEE.
Golding, P., & Donaldson, O. (2006). Predicting academic performance. In Frontiers in education
conference, 36th Annual (pp. 21–26): IEEE.
Goos, M., & Salomons, A. (2016). Measuring teaching quality in higher education: assessing selection
bias in course evaluations. Research in Higher Education, 58(4), 341–364.
Gowda, S.M., Baker, R.S., Corbett, A.T., Rossi, L.M. (2013). Towards automatically detecting whether
student learning is shallow. International Journal of Artificial Intelligence in Education, 23(1–4),
50–70.
Gray, G., McGuinness, C., Owende, P. (2014). An application of classification models to predict
learner progression in tertiary education. In Advance Computing Conference (IACC), 2014 IEEE
International (pp. 549–554): IEEE.
Grivokostopoulou, F., Perikos, I., Hatzilygeroudis, I. (2014). Utilizing semantic web technologies and
data mining techniques to analyze students learning and predict final performance. In 2014 IEEE
international conference on teaching, assessment and learning for engineering (TALE) (pp. 488–494):
IEEE.
Guarin, C.E.L., Guzman, E.L., Gonzalez, F.A. (2015). A model to predict low academic performance at a
specific enrollment using data mining. Revista Iberoamericana de Tecnologias del Aprendizaje, 10(3),
119–125.
Guo, B., Zhang, R., Xu, G., Shi, C., Yang, L. (2015). Predicting students performance in educational data
mining. In International symposium on educational technology, ISET 2015 (pp. 125–128).
Guruler, H., Istanbullu, A., Karahasan, M. (2010). A new student performance analysing system using
knowledge discovery in higher educational databases. Computers & Education, 55(1), 247–254.
Hämäläinen, W., & Vinni, M. (2006). Comparison of machine learning methods for intelligent tutoring
systems. In International conference on intelligent tutoring systems (pp. 525–534): Springer.
Hart, S., Daucourt, M., Ganley, C. (2017). Individual differences related to college students’ course
performance in calculus II. Journal of Learning Analytics, 4(2), 129–153.
Hasan, R., Palaniappan, S., Raziff, A.R.A., Mahmood, S., Sarker, K.U. (2018). Student academic perfor-
mance prediction by using decision tree algorithm. In 2018 4th international conference on computer
and information sciences (ICCOINS) (pp. 1–5): IEEE.
Hasheminejad, H., & Sarvmili, M. (2018). S3PSO: students’ performance prediction based on particle
swarm optimization. Journal of AI and Data Mining, 7(1), 77–96.
Hassan, O.R., & Rasiah, R. (2011). Poverty and student performance in Malaysia. International Journal
of Institutions and Economies, 3(1), 61–76.
Hattie, J., & Clinton, J. (2012). Physical activity is not related to performance at school. Archives of
Pediatrics & Adolescent Medicine, 166(7), 678–679.
Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–
112.
Helal, S., Li, J., Liu, L., Ebrahimie, E., Dawson, S., Murray, D.J. (2018). Identifying key factors of student
academic performance by subgroup discovery. International Journal of Data Science and Analytics,
7(3), 227–245.
Hidayah, I., Permanasari, A.E., Ratwastuti, N. (2013). Student classification for academic performance
prediction using neuro fuzzy in a conventional classroom. In 2013 international conference on
information technology and electrical engineering (ICITEE) (pp. 221–225): IEEE.
Hong, B., Wei, Z., Yang, Y. (2017). Online education performance prediction via time-related features. In
2017 IEEE/ACIS 16th international conference on computer and information science (ICIS) (pp. 95–
100): IEEE.
Hsiao, I.H., & Lin, Y.L. (2017). Enriching programming content semantics: an evaluation of visual
analytics approach. Computers in Human Behavior, 72, 771–782.
Hsiao, I.H., Pandhalkudi Govindarajan, S.K., Lin, Y.L. (2016). Semantic visual analytics for today’s pro-
gramming courses. In Proceedings of the sixth international conference on learning analytics and
knowledge (pp. 48–53): ACM.
Hu, X., Cheong, C.W.L., Ding, W., Woo, M. (2017). A systematic review of studies on predicting stu-
dent learning outcomes using learning analytics. In Proceedings of the seventh international learning
analytics & knowledge conference (pp. 528–529): ACM.
Huang, S., & Fang, N. (2013). Predicting student academic performance in an engineering dynamics
course: a comparison of four types of predictive mathematical models. Computers & Education, 61,
133–145.
Ibrahim, Z., & Rusli, D. (2007). Predicting students’ academic performance: comparing artificial neural
network, decision tree and linear regression. In 21st Annual SAS Malaysia Forum (pp. 1–6).
Ivančević, V., Čeliković, M., Luković, I. (2010). Analyzing student spatial deployment in a computer
laboratory. In Educational data mining (p. 2011).
Jara, M., & Mellar, H. (2010). Quality enhancement for e-learning courses: the role of student feedback.
Computers & Education, 54(3), 709–714.
Jishan, S.T., Rashu, R.I., Haque, N., Rahman, R.M. (2015). Improving accuracy of students’ final grade
prediction model using optimal equal width binning and synthetic minority over-sampling technique.
Decision Analytics, 2(1), 1.
Kabakchieva, D. (2012). Student performance prediction by using data mining classification algorithms.
International Journal of Computer Science and Management Research, 1(4), 686–690.
Kabra, R.R., & Bichkar, R.S. (2011). Performance prediction of engineering students using decision trees.
International Journal of Computer Applications, 36(11), 975–8887.
Kagdi, H., Collard, M.L., Maletic, J.I. (2007). A survey and taxonomy of approaches for mining software
repositories in the context of software evolution. Journal of Software Maintenance and Evolution:
Research and Practice, 19(2), 77–131.
Kamley, S., Jaloree, S., Thakur, R.S. (2016). A review and performance prediction of students’ using
association rule mining based approach. Data Mining and Knowledge Engineering, 8(8), 252–259.
Kaufman, L., & Rousseeuw, P.J. (2009). Finding groups in data: an introduction to cluster analysis
Vol. 344. New York: Wiley.
Kaviyarasi, R., & Balasubramanian, T. (2018). Exploring the high potential factors that affects students’
academic performance. International Journal of Education and Management Engineering, 8(6), 15.
Kesavaraj, G., & Sukumaran, S. (2013). A study on classification techniques in data mining. In 2013 fourth
international conference on computing, communications and networking technologies (ICCCNT)
(pp. 1–7): IEEE.
Khan, A., & Ghosh, S.K. (2016). Analysing the impact of poor teaching on student performance. In
2016 IEEE international conference on teaching, assessment, and learning for engineering (TALE)
(pp. 169–175): IEEE.
Khan, A., & Ghosh, S.K. (2018). Data mining based analysis to explore the effect of teaching on student
performance. Education and Information Technologies, 23(4), 1677–1697.
Khanna, L., Singh, S.N., Alam, M. (2016). Educational data mining and its role in determining fac-
tors affecting students academic performance: a systematic review. In 2016 1st India international
conference on information processing (IICIP) (pp. 1–7): IEEE.
Kim, B.H., Vizitei, E., Ganapathi, V. (2018). GritNet: student performance prediction with deep learning.
arXiv:1804.07405.
Koedinger, K.R., D’Mello, S., McLaughlin, E.A., Pardos, Z.A., Rosé, C.P. (2015). Data mining and
education. Wiley Interdisciplinary Reviews: Cognitive Science, 6(4), 333–353.
Kotsiantis, S.B., & Pintelas, P.E. (2005). Predicting students marks in hellenic open university. In Fifth
IEEE international conference on advanced learning technologies, 2005. ICALT 2005 (pp. 664–668):
IEEE.
Koutina, M., & Kermanidis, K.L. (2011). Predicting postgraduate students’ performance using machine
learning techniques. In IFIP international conference on artificial intelligence applications and
innovations (pp. 159–168): Springer.
Kumar, D.A., Selvam, R.P., Kumar, K.S. (2018). Review on prediction algorithms in educational data
mining. International Journal of Pure and Applied Mathematics, 118(8), 531–537.
Kumar, M., Singh, A.J., Handa, D. (2017). Literature survey on student’s performance prediction in educa-
tion using data mining techniques. International Journal of Education and Management Engineering,
6, 40–49.
Li, K.F., Rusk, D., Song, F. (2013). Predicting student academic performance. In 2013 seventh inter-
national conference on complex, intelligent, and software intensive systems (cisis) (pp. 27–33):
IEEE.
Lin, C.H., Kwon, J.B., Zhang, Y. (2018). Online self-paced high-school class size and student achievement.
Educational Technology Research and Development, pp 1–20.
Lipovetsky, S., & Conklin, W.M. (2015). Predictor relative importance and matching regression parame-
ters. Journal of Applied Statistics, 42(5), 1017–1031.
Liu, Q., Huang, Z., Huang, Z., Liu, C., Chen, E., Su, Y., et al. (2018a). Finding similar exercises in online
education systems. In Proceedings of the 24th ACM SIGKDD international conference on knowledge
discovery and data mining (pp. 1821–1830): ACM.
Liu, Q., Wu, R., Chen, E., Xu, G., Su, Y., Chen, Z., et al. (2018b). Fuzzy cognitive diagnosis for modelling
examinee performance. ACM Transactions on Intelligent Systems and Technology, 9(4), 48.
Livieris, I.E., Drakopoulou, K., Mikropoulos, T.A., Tampakas, V., Pintelas, P. (2018). An ensemble-based
semi-supervised approach for predicting students’ performance. In Research on e-Learning and ICT
in Education (pp. 25–42): Springer.
Loh, C.S., & Sheng, Y. (2015). Measuring the (dis-)similarity between expert and novice behaviors as
serious games analytics. Education and Information Technologies, 20(1), 5–19.
Lorenzetti, C., Maguitman, A., Leake, D., Menczer, F., Reichherzer, T. (2016). Mining for topics to suggest
knowledge model extensions. ACM Transactions on Knowledge Discovery from Data, 11(2), 23.
Lu, O.H.T., Huang, A.Y.Q., Huang, J.C., Lin, A.J.Q., Ogata, H., Yang, S.J.H. (2018). Applying learning
analytics for the early prediction of students’ academic performance in blended learning. Journal of
Educational Technology and Society, 21(2), 220–232.
Macfadyen, L.P., Dawson, S., Prest, S., Gašević, D. (2015). Whose feedback? A multilevel analysis of stu-
dent completion of end-of-term teaching evaluations. Assessment & Evaluation in Higher Education,
41(6), 821–839.
Márquez-Vera, C., Cano, A., Romero, C., Ventura, S. (2013). Predicting student failure at school using
genetic programming and different data mining approaches with high dimensional and imbalanced
data. Applied Intelligence, 38(3), 315–330.
Marsh, H.W. (1984). Students’ evaluations of university teaching: dimensionality, reliability, validity,
potential baises, and utility. Journal of Educational Psychology, 76(5), 707.
Marsh, H.W. (2007). Students’ evaluations of university teaching: Dimensionality, reliability, validity,
potential biases and usefulness. In The scholarship of teaching and learning in higher education: An
evidence-based perspective (pp. 319–383): Springer.
Martinez, D. (2001). Predicting Student Outcomes Using Discriminant Function Analysis.
Mat, U.B., Buniyamin, N., Arsad, P.M., Kassim, R. (2013). An overview of using academic analytics
to predict and improve students’ achievement: a proposed proactive intelligent intervention. In 2013
IEEE 5th conference on engineering education (ICEED) (pp. 126–130): IEEE.
McCarthy, K.S., & Goldman, S.R. (2017). Constructing interpretive inferences about literary text: the role
of domain-specific knowledge. Learning and Instruction, 60, 245–251.
Meier, Y., Xu, J., Atan, O., van der Schaar, M. (2016). Predicting grades. IEEE Transactions on Signal
Processing, 64(4), 959–972.
Miguéis, V., Freitas, A., Garcia, P.J., Silva, A. (2018). Early segmentation of students according to their
academic performance: a predictive modelling approach. Decision Support Systems, 115, 36–51.
Mimis, M., El Hajji, M., Es-saady, Y., Oueld Guejdi, A., Douzi, H., Mammass, D. (2018). A framework
for smart academic guidance using educational data mining. Education and Information Technologies,
24(2), 1379–1393.
Mishra, T., Kumar, D., Gupta, S. (2014). Mining students’ data for prediction performance. In Interna-
tional conference on advanced computing and communication technologies, ACCT (pp. 255–262).
Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G. (2009). Preferred reporting items for systematic reviews
and meta-analyses: the PRISMA statement. Annals of Internal Medicine, 151(4), 264–269.
Montuschi, P., Lamberti, F., Gatteschi, V., Demartini, C. (2015). A semantic recommender system for
adaptive learning. IT Professional, 17(5), 50–58.
Moore, S., & Kuol, N. (2005). Students evaluating teachers: exploring the importance of faculty reaction
to feedback on teaching. Teaching in Higher Education, 10(1), 57–73.
Moos, D.C., & Azevedo, R. (2008). Self-regulated learning with hypermedia: the role of prior domain
knowledge. Contemporary Educational Psychology, 33(2), 270–298.
Mueen, A., Zafar, B., Manzoor, U. (2016). Modeling and predicting students’ academic performance using
data mining techniques. International Journal of Modern Education and Computer Science, 8(11), 36.
Nakayama, M. (2016). Lexical analysis of syllabi in the area of technology enhanced learning. In
2016 15th international conference on information technology based higher education and training
(ITHET) (pp. 1–5): IEEE.
Natek, S., & Zwilling, M. (2014). Student data mining solution-knowledge management system related to
higher education institutions. Expert Systems with Applications, 41(14), 6400–6407.
Nikolic, S., Ritz, C., Vial, P.J., Ros, M., Stirling, D. (2015). Decoding student satisfaction: How to manage
and improve the laboratory experience. IEEE Transactions on Education, 58(3), 151–158.
O’Connell, K.A., Wostl, E., Crosslin, M., Berry, T.L., Grover, J.P. (2018). Student ability best predicts
final grade in a college algebra course. Journal of Learning Analytics, 5(3), 167–181.
Ogor, E.N. (2007). Student academic performance monitoring and evaluation using data mining tech-
niques. In Electronics, robotics and automotive mechanics conference (pp. 354–359): IEEE.
Ornelas, F., & Ordonez, C. (2017). Predicting student success: a naı̈ve Bayesian application to community
college data. Technology, Knowledge and Learning, 22(3), 299–315.
Osmanbegović, E., & Suljić, M. (2012). Data mining approach for predicting student performance.
Economic Review, 10(1), 3–12.
Ostrow, K., Donnelly, C., Heffernan, N. (2015). Optimizing partial credit algorithms to predict student
performance. In International conference on educational data mining (EDM).
Pal, S., & Chaurasia, V. (2017). Is alcohol affect higher education students performance: searching and
predicting pattern using data mining algorithms. International Journal of Innovations & Advancement
in Computer Science IJIACS ISSN, 6(4), 2347–8616.
Pandey, M., & Taruna, S. (2016). Towards the integration of multiple classifier pertaining to the Student’s
performance prediction. Perspectives in Science, 8, 364–366.
Pandey, U.K., & Pal, S. (2011). A data mining view on class room teaching language. arXiv:1104.4164.
Papamitsiou, Z.K., Terzis, V., Economides, A.A. (2014). Temporal learning analytics for computer based
testing. In Proceedings of the fourth international conference on learning analytics and knowledge
(pp. 31–35): ACM.
Parack, S., Zahid, Z., Merchant, F. (2012). Application of data mining in educational databases for pre-
dicting academic trends and patterns. In 2012 IEEE international conference on technology enhanced
education (ICTEE) (pp. 1–4): IEEE.
Pardos, Z.A., Heffernan, N.T., Anderson, B., Heffernan, C.L., Schools, W.P. (2010). Using fine-grained
skill models to fit student performance with Bayesian networks. Handbook of Educational Data
Mining, 417.
Pen̄a-Ayala, A. (2014). Educational data mining: a survey and a data mining-based analysis of recent
works. Expert Systems with Applications, 41(4), 1432–1462.
Polyzou, A., & Karypis, G. (2016). Grade prediction with course and student specific models. In Pacific-
Asia conference on knowledge discovery and data mining (pp. 89–101). Cham: Springer.
Polyzou, A., & Karypis, G. (2019). Feature extraction for next-term prediction of poor student perfor-
mance. IEEE Transactions on Learning Technologies, 12(2), 237–248.
Pong-Inwong, C., & Rungworawut, W. (2012). Teaching evaluation using data mining on moodle LMS
forum. In 2012 6th international conference on new trends in information science, service science and
data mining (ISSDM2012) (pp. 550–555): IEEE.
Price, L., Svensson, I., Borell, J., Richardson, J.T.E. (2017). The role of gender in students’ ratings
of teaching quality in computer science and environmental engineering. IEEE Transactions on
Education, 60(4), 281–287.
Quadri, M.M.N., & Kalyankar, N.V. (2010). Drop out feature of student data for academic performance
using decision tree techniques. Global Journal of Computer Science and Technology, 10(2).
Quille, K., & Bergin, S. (2018). Programming: Predicting student success early in CS1. A re-validation and
replication study. In Proceedings of the 23rd annual ACM conference on innovation and technology
in computer science education (pp. 15–20): ACM.
Quinlan, J.R. (1990). Decision trees and decision-making. IEEE Transactions on Systems, Man, and
Cybernetics, 20(2), 339–346.
Ramesh, V., Parkavi, P., Ramar, K. (2013). Predicting student performance: a statistical and data mining
approach. International Journal of Computer Applications, 63(8), 35–39.
Rani, S., & Kumar, P. (2017). A sentiment analysis system to improve teaching and learning. Computer,
50(5), 36–43.
Rekha, R., Angadi, A., Pathak, A., Kapur, A., Gosar, H., Ramanathan, M., et al. (2012). Ontology
driven framework for assessing the syllabus fairness of a question paper. In 2012 IEEE international
conference on technology enhanced education (ICTEE) (pp. 1–5): IEEE.
Romero, C., López, M.I., Luna, J.M., Ventura, S. (2013). Predicting students’ final performance from
participation in on-line discussion forums. Computers and Education, 68, 458–472.
Romero, C., & Ventura, S. (2007). Educational data mining: a survey from 1995 to 2005. Expert Systems
with Applications, 33(1), 135–146.
Romero, C., & Ventura, S. (2010). Educational data mining: a review of the state of the art. IEEE
Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews, 40(6), 601–618.
Romero, C., Ventura, S., Espejo, P.G., Hervás, C. (2008). Data mining algorithms to classify students. In
Educational data mining 2008.
Saarela, M., & Kärkkäinen, T. (2015). Analysing student performance using sparse data of core bachelor
courses. Journal of Educational Data Mining, 7(1), 3–32.
Sandoval, A., Gonzalez, C., Alarcon, R., Pichara, K., Montenegro, M. (2018). Centralized student perfor-
mance prediction in large courses based on low-cost variables in an institutional context. The Internet
and Higher Education, 37, 76–89.
Santana, M.A., Costa, E.B., Fonseca, B., Rego, J., de Araújo, F.F. (2017). Evaluating the effectiveness of
educational data mining techniques for early prediction of students’ academic failure in introductory
programming courses. Computers in Human Behavior, 73, 247–256.
Saxena, P.S., & Govil, M.C. (2009). Prediction of student’s academic performance using clustering. In
National conference on cloud computing & big data (pp. 1–6).
Shahiri, A.M., Husain, W., Rashid, A.N. (2015). A review on predicting student’s performance using data
mining techniques. Procedia Computer Science, 72, 414–422.
She, H.C., Cheng, M.T., Li, T.W., Wang, C.Y., Chiu, H.T., Lee, P.Z., et al. (2012). Web-based under-
graduate chemistry problem-solving: the interplay of task performance, domain knowledge and
web-searching strategies. Computers & Education, 59(2), 750–761.
Shingari, I., & Kumar, D. (2018). A survey on various aspects of education data mining in predicting
student performance. Journal of Applied Science and Computations, 5(6), 38–42.
Siemens, G., & Baker, R.S.J.D. (2012). Learning analytics and educational data mining: towards commu-
nication and collaboration. In Proceedings of the 2nd international conference on learning analytics
and knowledge (pp. 252–254): ACM.
Sivakumar, S., & Selvaraj, R. (2018). Predictive modeling of students performance through the enhanced
decision tree. In Advances in electronics, communication and computing (pp. 21–36). Singapore:
Springer.
Stronge, J.H., Ward, T.J., Tucker, P.D., Hindman, J.L. (2007). What is the relationship between teacher
quality and student achievement? An exploratory study. Journal of Personnel Evaluation in Educa-
tion, 20(3-4), 165–184.
Sullivan, W., Marr, J., Hu, G. (2017). A predictive model for standardized test performance in michigan
schools. In Applied computing and information technology (pp. 31–46): Springer.
Superby, J.F., Vandamme, J.P., Meskens, N. (2006). Determination of factors influencing the achievement
of the first-year university students using data mining methods. In Workshop on educational data
mining, (Vol. 32 p. 234).
Sweeney, M., Rangwala, H., Lester, J., Johri, A. (2016). Next-term student performance prediction: a
recommender systems approach. Journal of Educational Data Mining, 8(1), 22–51.
Tair, M.M.A., & El-Halees, A.M. (2012). Mining educational data to improve students’ performance: a
case study. International Journal of Information, 2(2), 140–146.
Thai-Nghe, N., Busche, A., Schmidt-Thieme, L. (2009). Improving academic performance prediction by
dealing with class imbalance. In Ninth international conference on intelligent systems design and
applications (pp. 878–883): IEEE.
Uddin, M.F., & Lee, J. (2017). Proposing stochastic probability-based math model and algorithms utilizing
social networking and academic data for good fit students prediction. Social Network Analysis and
Mining, 7(1), 29.
Üstünlüoğlu, E. (2016). Teaching quality matters in higher education: a case study from Turkey and
Slovakia. Teachers and Teaching, 23(3), 367–382.
Uttl, B., White, C.A., Gonzalez, D.W. (2017). Meta-analysis of faculty’s teaching effectiveness: Student
evaluation of teaching ratings and student learning are not related. Studies in Educational Evaluation,
54, 22–42.
Van Inwegen, E., Adjei, S., Wang, Y., Heffernan, N. (2015). An analysis of the impact of action order on
future performance: the fine-grain action model. In Proceedings of the fifth international conference
on learning analytics and knowledge (pp. 320–324): ACM.
VeeraManickam, M.R.M., Mohanapriya, M., Pandey, B.K., Akhade, S., Kale, S.A., Patil, R., et al. (2018).
Map-reduce framework based cluster architecture for academic student’s performance prediction
using cumulative dragonfly based neural network. Cluster Computing, 22(1), 1259–1275.
Walters, W.H. (2007). Google scholar coverage of a multidisciplinary field. Information Processing and
Management, 43(4), 1121–1132.
Wang, Y., Ostrow, K., Adjei, S., Heffernan, N. (2016). The opportunity count model: a flexible approach
to modeling student performance. In Proceedings of the Third (2016) ACM Conference on Learning@
Scale (pp. 113–116): ACM.
Widyahastuti, F., & Tjhin, V.U. (2017). Predicting students performance in final examination using linear
regression and multilayer perceptron. In 10th international conference on human system interactions
(HSI) (pp. 188–192): IEEE.
Willoughby, T., Anderson, S.A., Wood, E., Mueller, J., Ross, C. (2009). Fast searching for information on
the Internet to use in a learning context: the impact of domain knowledge. Computers & Education,
52(3), 640–648.
Wook, M., Yusof, Z.M., Nazri, M.Z.A. (2016). Educational data mining acceptance among undergraduate
students. Education and Information Technologies, 22(3), 1195–1216.
Xing, W., Guo, R., Petakovic, E., Goggins, S. (2015). Participation-based student final performance pre-
diction model through interpretable Genetic Programming: Integrating learning analytics, educational
data mining and theory. Computers in Human Behavior, 47, 168–181.
Xiong, X., Adjei, S., Heffernan, N. (2014). Improving retention performance prediction with prerequisite
skill features. In Educational data mining 2014.
Xu, M., Liang, Y., Wu, W. (2017). Predicting honors student performance using RBFNN and PCA method.
In International Conference on database systems for advanced applications (pp. 364–375): Springer.
Yang, S.J.H., Lu, O.H.T., Huang, A.Y.Q., Huang, J.C.H., Ogata, H., Lin, A.J.Q. (2018). Predicting stu-
dents’ academic performance using multiple linear regression and principal component analysis.
Journal of Information Processing, 26, 170–176.
Yim, S., & Warschauer, M. (2017). Web-based collaborative writing in L2 contexts: methodological
insights from text mining. Language Learning & Technology, 21(1), 146–165.
Yin, H., Wang, W., Han, J. (2016). Chinese undergraduates’ perceptions of teaching quality and the effects
on approaches to studying and course satisfaction. Higher Education, 71(1), 39–57.
Yoo, J., & Kim, J. (2014). Can online discussion participation predict group project performance? Inves-
tigating the roles of linguistic features and participation patterns. International Journal of Artificial
Intelligence in Education, 24(1), 8–32.
Yu, L., Lee, C., Pan, H., Chou, C., Chao, P., Chen, Z., et al. (2018). Improving early prediction of academic
failure using sentiment analysis on self evaluated comments. Journal of Computer Assisted Learning,
34(4), 358–365.
Zabaleta, F. (2007). The use and misuse of student evaluations of teaching. Teaching in Higher Education,
12(1), 55–76.
Zacharis, N.Z. (2015). A multivariate approach to predicting student outcomes in web-enabled blended
learning courses. The Internet and Higher Education, 27, 44–53.
Zaugg, H., West, R.E., Tateishi, I., Randall, D.L. (2011). Mendeley: Creating communities of scholarly
inquiry through research collaboration. TechTrends, 55(1), 32–36.
Zimmermann, J., Brodersen, K.H., Heinimann, H.R., Buhmann, J.M. (2015). A model-based approach to
predicting graduate-level performance using indicators of undergraduate-level performance. Journal
of Educational Data Mining, 7(3), 151–176.
Zhang, X., Sun, G., Pan, Y., Sun, H., He, Y., Tan, J. (2018). Students performance modeling based on
behavior pattern. Journal of Ambient Intelligence and Humanized Computing, 9(5), 1659–1670.
Zollanvari, A., Kizilirmak, R.C., Kho, Y.H., Hernández-Torrano, D. (2017). Predicting students’ GPA
and developing intervention strategies based on self-regulatory learning behaviors. IEEE Access, 5,
23792–23802.
Zorilla, M.E., Garcı́a-Saiz, D., Balcázar, J.L. (2010). Towards parameter-free data mining: mining
educational data with yacaree. In Educational data mining 2011.
Zuber, M. (2014). A survey of data mining techniques for social network analysis. International Journal
of Research in Computer Engineering & Electronics, 3(6), 1–8.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

Student Performance Analysis and Prediction in Classroom Learning: A Review of Educational Data Mining Studies

Uploaded by

Copyright:

Available Formats

Student Performance Analysis and Prediction in Classroom Learning: A Review of Educational Data Mining Studies

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Student Performance Analysis and Prediction in Classroom Learning: A Review of Educational Data Mining Studies

Uploaded by

Copyright:

Available Formats

Educational and Information Technologies (2021) 26:205–240

Student performance analysis and prediction

Anupam Khan1 · Soumya K. Ghosh1

Keywords Student performance · Classroom learning · Performance prediction ·

1 Department of Computer Science and Engineering, Indian Institute of Technology Kharagpur,

Table 1 Scope of the survey

Focus Research Outcome

Fig. 1 Methodology adapted for this survey

Fig. 2 Taxonomy of research directions

3 Student performance analysis

3.1 Inﬂuencing factor identiﬁcation

3.1.1 Student background and behaviour

The exploration of student demographics, behaviour, academic and social back-

Table 2 Student background and behaviour related studies

Study Explored predictor Method

Ahmed et al. (2014) Gender, residence type ARM

Table 3 Assessment related studies

Study Explored predictor Method

Ahmed et al. (2014) Internal and external assessment, attendance ARM

Table 4 Teaching related studies

Study Explored predictor Method

Abrami et al. (2007) SETE REG

Table 5 Domain knowledge related studies

Study Explored predictor Method

Adjei et al. (2016) Prerequisite skills REG

ARM: Association rule mining, CTR: Clustering, REG: Regression

Analysing the effect of teaching is a long-debated topic in literature. A majority of

3.1.4 Domain knowledge

3.2 Methods used for identifying the factors

Researchers have applied various techniques such as clustering, classification, asso-

3.2.1 Association rule mining

in student performance analysis literature (Bydžovská 2016; Campagni et al. 2015;

3.2.5 Text mining and social network analysis

4 Student performance prediction

et al. 2016; Márquez-Vera et al. 2013). In addition to efficiency, time of predic-

4.1 Time of prediction

4.1.1 During the tenure of the course

Table 6 Classification studies during the tenure of the course

Study Predictor used Aim Efficiency

Study Predictor used Aim Efficiency

Acc: Accuracy, AUC: Area under curve, FM: F-measure

Table 7 Regression studies during the tenure of the course

Study Predictor used Aim Efficiency

Cen et al. (2016) Student interaction Final grade MAE: 0.0800

Table 8 Classification studies before course commencement

Study Predictor used Aim Efficiency

Chanlekha and Niramitranon Student demographics, performance At-risk Acc: 62.50 %

Acc: Accuracy, RMSE: Root mean square error

Table 9 Regression studies before course commencement

Study Predictor used Aim Efficiency

4.1.2 Before course commencement

4.2 Aim of prediction

4.2.1 Success prediction

Modelling the actual grade or marks is a challenging task as it depends on diverse

4.2.2 Grade or marks prediction

Fig. 3 Year-wise number of studies

Table 10 Comparison of various predictors used in student performance prediction studies

No of studies * Avg. Accuracy

Predictor type After Before After Before